diff --git a/.specify/feature.json b/.specify/feature.json
index d11bc89..f3600ea 100644
--- a/.specify/feature.json
+++ b/.specify/feature.json
@@ -1,3 +1,3 @@
 {
-  "feature_directory": "docs/specs/082-syft-sharded-sbom-plan"
+  "feature_directory": "docs/specs/083-tool-acquisition-guidance"
 }
diff --git a/AGENTS.md b/AGENTS.md
index c77bf9e..79553db 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -202,5 +202,5 @@ go run ./cmd/portolan scan --help
 <!-- SPECKIT START -->
 For additional context about technologies to be used, project structure,
 shell commands, and other important information, read the current plan:
-`docs/specs/082-syft-sharded-sbom-plan/plan.md`
+`docs/specs/083-tool-acquisition-guidance/plan.md`
 <!-- SPECKIT END -->
diff --git a/docs/product-backlog.md b/docs/product-backlog.md
index c30b380..06753ed 100644
--- a/docs/product-backlog.md
+++ b/docs/product-backlog.md
@@ -164,6 +164,7 @@ fixtures are preflight evidence only.
 | P6-080 | `docs/specs/080-clean-start-artifact-guard/` | Context packs and acceptance guidance make the current artifact boundary explicit so Cursor/agent stress lanes do not mix stale `.portolan/stress`, root-level `run`, or unrelated generated outputs into clean-start evidence. | Ready-for-review PR #58; post-Cursor local baseline, fresh Bigtop context smoke, final Cursor Composer 2.5 clean-start stress, and three final assessed non-GPT review lanes verified; merge approval `not_assessed` |
 | P6-081 | `docs/specs/081-maven-sharded-producer-plan/` | Context packs emit repository-sharded Maven/CycloneDX next actions for multi-repo JVM landscapes so agents do not treat one sample `pom.xml` as a landscape rollout plan. | Ready-for-review PR #59; local baseline, fresh Bigtop context smoke, Cursor Composer 2.5 stress, three assessed non-GPT review lanes, and GitHub checks verified. Maven execution, dependency evidence, JVM relationship claims, GitHub review approval, and merge approval remain `not_assessed` |
 | P6-082 | `docs/specs/082-syft-sharded-sbom-plan/` | Context packs emit repository-sharded Syft/CycloneDX SBOM next actions for multi-repo landscapes so component/dependency evidence can be acquired incrementally without full-root SBOM scans. | Ready-for-review PR #60; local baseline, fresh Bigtop context smoke, Cursor Composer 2.5 stress, three assessed non-GPT review lanes, and GitHub checks verified. Syft execution, component inventory, dependency evidence, GitHub review approval, and merge approval remain `not_assessed` |
+| P6-083 | `docs/specs/083-tool-acquisition-guidance/` | Context packs make tool acquisition guidance explicit and stack-agnostic: agents can pull in the right local producer tools without treating Portolan as a PHP/JVM/Gradle adapter stack. | Ready-for-review PR #61; local baseline, fresh Bigtop context smoke, Cursor Composer 2.5 stress, integrated PR #57-#61 stack-agnostic stress, three assessed non-GPT review lanes, and GitHub checks verified. Native producer execution, tool install/acquisition, component inventory, dependency relationships, duplication metrics, runtime topology, GitHub review approval, and merge approval remain `not_assessed` |
 
 ## Backlog Rules
 
diff --git a/docs/specs/083-tool-acquisition-guidance/plan.md b/docs/specs/083-tool-acquisition-guidance/plan.md
new file mode 100644
index 0000000..aca0a9b
--- /dev/null
+++ b/docs/specs/083-tool-acquisition-guidance/plan.md
@@ -0,0 +1,54 @@
+# Implementation Plan: Tool Acquisition Guidance
+
+**Branch**: `codex/083-tool-acquisition-guidance`
+
+**Date**: 2026-06-02
+
+**Spec**: `docs/specs/083-tool-acquisition-guidance/spec.md`
+
+## Summary
+
+Clarify the producer-planning surface so Portolan stays stack-agnostic while
+still telling agents which local tools to pull in for missing evidence families.
+The slice adds generic acquisition/risk guidance around existing OSS tool plans;
+it does not add a new scanner, install tools, run native producers, or create a
+language-specific adapter.
+
+## Decision Gate
+
+- **Simpler/Faster**: Leave the existing `oss-plan.json` text as-is. Rejected
+  because Cursor and operator discussion showed the guidance can be misread as
+  incremental stack-specific adapter work.
+- **Blocking Edge Cases**: External tools can install packages, hit networks,
+  write caches, mutate targets, or expose private dependency coordinates.
+  Therefore acquisition guidance is descriptive and approval-gated; no native
+  command is run by Portolan.
+- **Existing Open Source**: Continue composing mature OSS/native producer tools
+  such as Syft, CycloneDX plugins, jscpd, and Semgrep. Portolan owns the
+  evidence contract and normalization boundary, not their scanner logic.
+
+## Technical Context
+
+**Language/Version**: Go.
+
+**Primary Dependencies**: Standard library only. No new dependency.
+
+**Storage**: Local context artifacts and SpecKit docs.
+
+**Testing**: Focused `internal/contextprep` test, then full baseline.
+
+**Constraints**: Local-first/read-only defaults. No installs, network access,
+producer execution, daemon behavior, credentials, or target mutation.
+
+## Verification
+
+```bash
+go test ./internal/contextprep
+go test ./...
+go vet ./...
+jq empty schema/*.json
+git diff --check
+```
+
+Fresh Bigtop smoke and Cursor Composer 2.5 stress must use a clean context path
+and must not run native producers.
diff --git a/docs/specs/083-tool-acquisition-guidance/research.md b/docs/specs/083-tool-acquisition-guidance/research.md
new file mode 100644
index 0000000..5e21bde
--- /dev/null
+++ b/docs/specs/083-tool-acquisition-guidance/research.md
@@ -0,0 +1,25 @@
+# Research: Tool Acquisition Guidance
+
+## Decision: Generic acquisition guidance over per-stack adapters
+
+- **Decision**: Keep producer planning organized by evidence family and local
+  tool candidate, not by programming-language adapter.
+- **Rationale**: The operator needs practical next actions such as "pull in a
+  local SBOM producer" or "run a duplication tool", but Portolan must not become
+  a JVM/PHP/Gradle scanner.
+- **Rejected alternative**: Add a Gradle-specific slice after integrated stress.
+  Rejected because it turns a residual tool gap into stack ownership.
+
+## OSS posture
+
+- Mature OSS/native tools remain the right acquisition targets when they
+  produce local files that Portolan can normalize.
+- Tool recommendations are options, not evidence.
+- Missing or unrun candidate tools remain `not_assessed`.
+
+## Risk posture
+
+- Tool acquisition can involve network access, cache writes, project mutation,
+  dependency coordinate exposure, or runtime side effects.
+- Portolan must surface those risks before suggesting any command as a next
+  action.
diff --git a/docs/specs/083-tool-acquisition-guidance/reviews/bigtop-context-and-cursor-stress-2026-06-02.md b/docs/specs/083-tool-acquisition-guidance/reviews/bigtop-context-and-cursor-stress-2026-06-02.md
new file mode 100644
index 0000000..6f0abc8
--- /dev/null
+++ b/docs/specs/083-tool-acquisition-guidance/reviews/bigtop-context-and-cursor-stress-2026-06-02.md
@@ -0,0 +1,69 @@
+# Bigtop Context Smoke And Cursor Stress
+
+Date: 2026-06-02
+
+Spec: `docs/specs/083-tool-acquisition-guidance/`
+
+## Fresh Context Smoke
+
+Command:
+
+```bash
+go run ./cmd/portolan context prepare \
+  --root /home/fall_out_bug/projects/bigtop-landscape \
+  --out /home/fall_out_bug/projects/bigtop-landscape/.portolan/stress/20260602-083-tool-acquisition-guidance/context \
+  --profile cursor \
+  --force
+```
+
+verified:
+
+- Context pack was written under:
+  `/home/fall_out_bug/projects/bigtop-landscape/.portolan/stress/20260602-083-tool-acquisition-guidance/context`
+- `context/tool-outputs` is absent; no native producer was executed.
+- JSON validation passed for `repos.json`, `tool-registry.json`,
+  `oss-plan.json`, and `gaps.jsonl`.
+- Selected tool acquisition states:
+  - `cyclonedx`: installed / `not_assessed`
+  - `jscpd`: installed / `not_assessed`
+  - `maven-cyclonedx`: installed / `not_assessed`
+  - `gradle-cyclonedx`: installed but requires local evaluation /
+    `not_assessed`
+
+not_assessed:
+
+- Actual Syft, jscpd, Maven, Gradle, Semgrep, Docker, or producer execution.
+- Actual producer output validity.
+- Component inventory, dependency relationships, duplication metrics, and
+  runtime topology.
+
+## Cursor Composer 2.5 Stress
+
+Command:
+
+```bash
+cursor-agent --print --mode ask --model composer-2.5 --trust "$(cat docs/specs/083-tool-acquisition-guidance/stress/cursor-tool-acquisition-prompt-2026-06-02.md)"
+```
+
+verified:
+
+- `forbidden_read: false`
+- `artifacts_read_count: 8`
+- `acquisition_guidance_present: true`
+- `acquisition_tool_count: 5`
+- `stack_specific_adapter_requested: false`
+- `candidate_tools_as_evidence: false`
+- `installed_tools_claimed_as_supported_evidence: false`
+- `approval_boundary_present: true`
+- `risks_named: true`
+- `evidence_until_output: not_assessed`
+- `component_dependency_claimable: false`
+- `runtime_topology_claimable: false`
+- `verdict: pass`
+
+disposition:
+
+- Accepted as a passing stress lane for the tool acquisition guidance
+  correction.
+- Residual clean-start producer-run handling is owned by pending PR #58 and was
+  already verified in the integrated PR #57-#60 scratch stress.
diff --git a/docs/specs/083-tool-acquisition-guidance/reviews/pr-readiness-closeout-2026-06-02.md b/docs/specs/083-tool-acquisition-guidance/reviews/pr-readiness-closeout-2026-06-02.md
new file mode 100644
index 0000000..7d98e31
--- /dev/null
+++ b/docs/specs/083-tool-acquisition-guidance/reviews/pr-readiness-closeout-2026-06-02.md
@@ -0,0 +1,91 @@
+# PR Readiness Closeout
+
+Date: 2026-06-02
+
+Spec: `docs/specs/083-tool-acquisition-guidance/`
+
+PR: https://github.com/fcon-tech/portolan/pull/61
+
+Branch: `codex/083-tool-acquisition-guidance`
+
+Head at PR creation: `da2b2072b1c7decdc7954327d8f0a29b3bc4fbe6`
+
+## Implementation State
+
+verified:
+
+- `oss-plan.json` tool plans include stack-agnostic acquisition guidance.
+- Candidate tools are represented as native producer options, not
+  Portolan-owned stack adapters.
+- Tool availability remains separate from evidence.
+- `evidence_until_output` remains `not_assessed` until local output is produced
+  and re-ingested.
+- Answer and query guidance reject defaulting to PHP/JVM/Scala/Gradle adapter
+  requests.
+
+not_assessed:
+
+- Actual native producer execution.
+- Actual tool install/acquisition.
+- Component inventory, dependency relationships, duplication metrics, and
+  runtime topology.
+
+## Local Verification
+
+verified:
+
+- `go test ./internal/contextprep`
+- `go test ./...`
+- `go vet ./...`
+- `jq empty schema/*.json`
+- `git diff --check`
+- Fresh Bigtop context smoke.
+- Cursor Agent `composer-2.5` bounded tool-acquisition stress.
+
+## Review Evidence
+
+verified:
+
+- Requirements/product-vision drift review recorded.
+- Cursor Composer 2.5 stress recorded.
+- Three assessed non-GPT review lanes recorded:
+  - `openrouter/moonshotai/kimi-k2.6`
+  - `openrouter/deepseek/deepseek-v4-pro`
+  - `openrouter/qwen/qwen3-coder`
+- Degraded MiMo/MiniMax attempts recorded as non-counting evidence.
+
+## PR State
+
+verified at PR creation:
+
+- PR #61 exists.
+- PR is open.
+- PR is not draft.
+- PR head branch is `codex/083-tool-acquisition-guidance`.
+
+not_assessed at PR creation:
+
+- GitHub checks were queued.
+- GitHub review approval absent/not_assessed.
+- Ready-to-merge approval absent/not_assessed.
+
+verified after current-head refresh:
+
+- Current PR head: `bd6e15fad6981b15c14975e59a828c1f364da5f3`
+- PR is open and not draft.
+- `mergeStateStatus=CLEAN`.
+- Current GitHub checks: all reported checks completed successfully.
+- Integrated PR #57-#61 stack-agnostic stress is recorded under
+  `docs/specs/083-tool-acquisition-guidance/reviews/integrated-stack-agnostic-navigation-stress-2026-06-02.md`
+  on the scratch integration branch.
+
+## Readiness
+
+- Ready-for-review PR: yes.
+- Ready-to-merge PR: no.
+
+Stop reason:
+
+- PR is ready for review.
+- Do not merge without explicit user approval and current merge-state/check
+  verification.
diff --git a/docs/specs/083-tool-acquisition-guidance/reviews/requirements-product-vision-drift-2026-06-02.md b/docs/specs/083-tool-acquisition-guidance/reviews/requirements-product-vision-drift-2026-06-02.md
new file mode 100644
index 0000000..c3a57c8
--- /dev/null
+++ b/docs/specs/083-tool-acquisition-guidance/reviews/requirements-product-vision-drift-2026-06-02.md
@@ -0,0 +1,42 @@
+# Requirements And Product-Vision Drift Review
+
+Date: 2026-06-02
+
+Spec: `docs/specs/083-tool-acquisition-guidance/`
+
+## Inputs
+
+- Integrated Cursor Composer 2.5 stress for PR #57-#60.
+- User correction: Portolan must be stack-agnostic, but should help pull in the
+  right tools.
+- Portolan product boundary in `AGENTS.md`.
+- Constitution local-first, evidence-state, and OSS-composition principles.
+
+## Drift Assessment
+
+Requirements:
+
+- Aligned. The slice improves the navigation surface for missing evidence
+  families without adding a stack-specific scanner or adapter.
+
+Product boundary:
+
+- Aligned. Portolan remains a read-only local discovery substrate and
+  normalizer for local producer evidence.
+- The slice explicitly rejects defaulting to PHP/JVM/Gradle adapters.
+- Tool candidates remain options for local acquisition, not observed evidence.
+
+Evidence semantics:
+
+- Aligned. Candidate tools, install suggestions, and approval-gated commands
+  remain `not_assessed` until local output exists and is re-ingested.
+
+Open-source posture:
+
+- Aligned. Mature OSS/native producer tools stay outside Portolan ownership;
+  Portolan documents how to acquire and normalize their outputs safely.
+
+Decision:
+
+- Proceed with a focused context guidance correction.
+- Do not create a Gradle-specific implementation slice.
diff --git a/docs/specs/083-tool-acquisition-guidance/reviews/slice-review-disposition-2026-06-02.md b/docs/specs/083-tool-acquisition-guidance/reviews/slice-review-disposition-2026-06-02.md
new file mode 100644
index 0000000..49eafeb
--- /dev/null
+++ b/docs/specs/083-tool-acquisition-guidance/reviews/slice-review-disposition-2026-06-02.md
@@ -0,0 +1,108 @@
+# Slice Review Disposition
+
+Date: 2026-06-02
+
+Spec: `docs/specs/083-tool-acquisition-guidance/`
+
+## Independent Lanes
+
+| Lane | Model | State | Verdict |
+| --- | --- | --- | --- |
+| Kimi | `openrouter/moonshotai/kimi-k2.6` | assessed with degraded process text | Found no stack-boundary or evidence-state blockers. Raised schema-contract concern that was locally checked. |
+| DeepSeek | `openrouter/deepseek/deepseek-v4-pro` | assessed | Stack-agnostic boundary, evidence honesty, and local-first behavior pass. Raised backward-compat concern for old `oss-plan.json`. |
+| Qwen | `openrouter/qwen/qwen3-coder` | assessed | No actionable findings. |
+
+Degraded lanes:
+
+- `openrouter/xiaomi/mimo-v2.5-pro`: returned a tool-call request instead of a
+  review verdict; `not_assessed`.
+- `minimax/MiniMax-M2.7`: provider id returned 404; `failed`.
+- `openrouter/minimax/minimax-m2.7`: endpoint rejected no-reasoning mode;
+  `failed`.
+
+## Findings
+
+### F1: `oss-plan.json` schema-contract drift
+
+Sources: Kimi, DeepSeek
+
+Classification: major from reviewers; rejected as current blocker after local
+verification.
+
+Local verification:
+
+- `schema/` exists, but there is no committed `schema/oss-plan.json` contract in
+  current `origin/main`.
+- Current verification command `jq empty schema/*.json` checks committed schema
+  syntax only; it does not validate `oss-plan.json` output shape.
+- Current code writes `oss-plan.json` but does not consume older context packs
+  as a public input contract.
+
+Disposition:
+
+- Rejected as a blocker for this slice.
+- Recorded as a future contract-hardening option if `oss-plan.json` becomes a
+  formal schema-validated artifact.
+- The field is additive and does not remove existing JSON fields.
+
+### F2: Older `oss-plan.json` without `acquisition` may unmarshal with empty fields
+
+Source: DeepSeek
+
+Classification: major from reviewer; rejected as current blocker.
+
+Rationale:
+
+- No current Portolan code path reads old `oss-plan.json` as an input contract.
+- The new field is generated for all current tool plans.
+- Downstream consumers that ignore unknown fields remain compatible; consumers
+  requiring a formal schema need a future `oss-plan` schema/contract slice.
+
+### F3: Availability-string coverage could be stricter
+
+Source: DeepSeek
+
+Classification: minor
+
+Disposition: non-blocking.
+
+Rationale:
+
+- Focused test already verifies Gradle's local-evaluation next action and
+  acquisition presence.
+- Availability exact strings are less important than evidence-state and
+  approval boundaries.
+
+### F4: Gradle/Semgrep remain partial next actions
+
+Sources: Cursor, DeepSeek
+
+Classification: residual, not a 083 defect.
+
+Disposition: accepted as residual.
+
+Rationale:
+
+- 083 makes acquisition guidance explicit; it does not add new bounded native
+  commands.
+- `not_assessed` plus local-evaluation guidance is the desired boundary for
+  tools without safe output-bounded commands.
+
+## Verification
+
+verified:
+
+- `go test ./internal/contextprep`
+- `go test ./...`
+- `go vet ./...`
+- `jq empty schema/*.json`
+- `git diff --check`
+- Fresh Bigtop context smoke.
+- Cursor Agent `composer-2.5` bounded tool-acquisition stress.
+
+not_assessed:
+
+- Actual native producer execution.
+- Actual tool install/acquisition.
+- Formal JSON Schema validation for `oss-plan.json`, because no such committed
+  schema exists in the current repository.
diff --git a/docs/specs/083-tool-acquisition-guidance/spec.md b/docs/specs/083-tool-acquisition-guidance/spec.md
new file mode 100644
index 0000000..c470867
--- /dev/null
+++ b/docs/specs/083-tool-acquisition-guidance/spec.md
@@ -0,0 +1,60 @@
+# Feature Specification: Tool Acquisition Guidance
+
+**Feature Branch**: `codex/083-tool-acquisition-guidance`
+
+**Created**: 2026-06-02
+
+**Status**: Ready-for-review PR #61; local baseline, fresh Bigtop context
+smoke, Cursor Composer 2.5 stress, integrated PR #57-#61 stack-agnostic stress,
+three assessed non-GPT review lanes, and GitHub checks verified; merge approval
+`not_assessed`
+
+**Input**: Integrated Cursor Composer 2.5 stress for PR #57-#60 judged the
+context adequate as a pre-UX navigation harness, but the follow-up discussion
+clarified the product boundary: Portolan must stay stack-agnostic while still
+helping the operator pull in the right local producer tools.
+
+## User Scenarios & Testing
+
+### User Story 1 - Stack-Agnostic Tool Acquisition (Priority: P1)
+
+An agent reading a context pack sees missing evidence families, candidate local
+producer tools, availability, approval/risk boundaries, and next actions
+without interpreting them as Portolan-owned language adapters.
+
+**Independent Test**: A fixture with Maven/Gradle manifests and local tools
+shows `oss-plan.json` acquisition guidance for native producer candidates while
+the answer contract says candidate tools are not Portolan adapters.
+
+### User Story 2 - Evidence Honesty (Priority: P1)
+
+An agent can recommend installing or running a local tool only as a way to
+produce future local evidence; it must keep the evidence family `not_assessed`
+until output exists and is re-ingested.
+
+**Independent Test**: Context guidance includes `not_assessed` boundaries for
+candidate tools and explicitly requires context refresh after approved output.
+
+## Requirements
+
+- **FR-001**: `oss-plan.json` MUST describe producer candidates as tool
+  acquisition guidance, not stack-specific Portolan adapters.
+- **FR-002**: Candidate tools MUST expose availability and a next acquisition
+  action such as run approved local command, install/evaluate locally, or supply
+  local output.
+- **FR-003**: Candidate tools MUST expose risk boundaries for network, target
+  mutation, cache writes, and approval where relevant.
+- **FR-004**: Guidance MUST preserve `not_assessed` evidence state until local
+  output exists and `context prepare --force` re-ingests it.
+- **FR-005**: Answer guidance MUST explicitly reject defaulting to
+  Portolan-owned PHP/JVM/Scala/Gradle adapters for language coverage gaps.
+
+## Success Criteria
+
+- **SC-001**: Focused contextprep tests verify acquisition guidance in
+  `oss-plan.json`, `answer-contract.md`, and `query-plan.md`.
+- **SC-002**: Fresh Bigtop context shows tool acquisition guidance without
+  running native producers.
+- **SC-003**: Cursor Composer 2.5 confirms the context remains stack-agnostic
+  while still surfacing useful local tools to pull in.
+- **SC-004**: Baseline checks pass.
diff --git a/docs/specs/083-tool-acquisition-guidance/stress/cursor-tool-acquisition-output-2026-06-02.md b/docs/specs/083-tool-acquisition-guidance/stress/cursor-tool-acquisition-output-2026-06-02.md
new file mode 100644
index 0000000..186aae9
--- /dev/null
+++ b/docs/specs/083-tool-acquisition-guidance/stress/cursor-tool-acquisition-output-2026-06-02.md
@@ -0,0 +1,61 @@
+# Portolan 083 Stress Evaluation: Tool Acquisition Guidance
+
+Assessment from allowed context only:
+
+```text
+/home/fall_out_bug/projects/bigtop-landscape/.portolan/stress/20260602-083-tool-acquisition-guidance/context
+```
+
+No forbidden paths were read and no producers were run.
+
+| Field | Value |
+| --- | --- |
+| `lane_state` | `completed` |
+| `artifacts_read_count` | 8 |
+| `forbidden_read` | false |
+| `acquisition_guidance_present` | true |
+| `acquisition_tool_count` | 5 |
+| `stack_specific_adapter_requested` | false |
+| `candidate_tools_as_evidence` | false |
+| `installed_tools_claimed_as_supported_evidence` | false |
+| `approval_boundary_present` | true |
+| `risks_named` | true |
+| `evidence_until_output` | `not_assessed` |
+| `producer_outputs_present` | partial |
+| `component_dependency_claimable` | false |
+| `runtime_topology_claimable` | false |
+| `next_actions_specific_enough` | true for Syft, jscpd, and Maven; partial for Gradle/Semgrep |
+| `verdict` | pass |
+
+Cursor verdict:
+
+> Portolan behaves as a stack-agnostic local-first navigation harness: it routes
+> agents to approval-gated native producers without posing them as
+> Portolan-owned language adapters, and keeps producer/candidate evidence
+> `not_assessed` until local output is produced and re-ingested.
+
+## Key Evidence
+
+- `oss-plan.json` includes `acquisition` guidance on all 5 tool plans.
+- Candidate tools are local producer options, not PHP/JVM/Scala/Gradle
+  adapters.
+- `producer-recommendation` records remain `candidate_only` /
+  `not_assessed`.
+- Installed Syft, jscpd, and Maven availability does not become evidence:
+  `evidence_state` and `evidence_until_output` remain `not_assessed`.
+- Approval, network, mutation, and output-not-evidence-until-reingested risks
+  are named.
+- Component/dependency and runtime-topology claims remain blocked.
+
+## Residual Gaps
+
+- Empty `tool-registry.json`.
+- No map bundle in context.
+- Native map remains limited to Go.
+- Gradle/Semgrep have no bounded command in this isolated branch.
+- This branch is based on `origin/main` and does not include pending clean-start
+  guard PR #58, so older verified producer-run records can still appear as
+  metadata-visible prior proof paths. That is not a 083 regression, but the
+  integrated PR #57-#60 scratch stress already showed the clean-start guard
+  scrubs sibling stress outputs when #58 is included.
+- External completeness remains `unknown`.
diff --git a/docs/specs/083-tool-acquisition-guidance/stress/cursor-tool-acquisition-prompt-2026-06-02.md b/docs/specs/083-tool-acquisition-guidance/stress/cursor-tool-acquisition-prompt-2026-06-02.md
new file mode 100644
index 0000000..ea2801e
--- /dev/null
+++ b/docs/specs/083-tool-acquisition-guidance/stress/cursor-tool-acquisition-prompt-2026-06-02.md
@@ -0,0 +1,42 @@
+# Cursor Composer 2.5 Stress Prompt: Tool Acquisition Guidance
+
+Evaluate Portolan as a stack-agnostic local-first navigation harness after a
+tool acquisition guidance correction.
+
+Allowed context only:
+`/home/fall_out_bug/projects/bigtop-landscape/.portolan/stress/20260602-083-tool-acquisition-guidance/context`
+
+Forbidden reads/actions:
+
+- do not read sibling `.portolan/stress/*` roots
+- do not read root `run/` or map bundles outside this context
+- do not read `/home/fall_out_bug/projects/bigtop-landscape/repos` source files
+- do not run Syft, Maven, Gradle, jscpd, Docker, or any producer
+- do not install tools or mutate the target
+
+Task:
+
+Judge whether Portolan stays stack-agnostic while still helping an agent pull
+in the right local producer tools. Verify that candidate tools are not treated
+as Portolan-owned PHP/JVM/Gradle adapters and that all producer evidence remains
+`not_assessed` until local output exists.
+
+Required output fields:
+
+- lane_state
+- artifacts_read_count
+- forbidden_read
+- acquisition_guidance_present
+- acquisition_tool_count
+- stack_specific_adapter_requested
+- candidate_tools_as_evidence
+- installed_tools_claimed_as_supported_evidence
+- approval_boundary_present
+- risks_named
+- evidence_until_output
+- producer_outputs_present
+- component_dependency_claimable
+- runtime_topology_claimable
+- next_actions_specific_enough
+- residual_gaps
+- verdict
diff --git a/docs/specs/083-tool-acquisition-guidance/tasks.md b/docs/specs/083-tool-acquisition-guidance/tasks.md
new file mode 100644
index 0000000..165586c
--- /dev/null
+++ b/docs/specs/083-tool-acquisition-guidance/tasks.md
@@ -0,0 +1,27 @@
+# Tasks: Tool Acquisition Guidance
+
+**Input**: `docs/specs/083-tool-acquisition-guidance/`
+
+## Phase 1: Setup And Review
+
+- [x] T001 Create dedicated branch/worktree for spec 083.
+- [x] T002 Create concrete `spec.md`, `plan.md`, `research.md`, and `tasks.md`.
+- [x] T003 Update `.specify/feature.json`, `AGENTS.md`, and
+  `docs/product-backlog.md` for spec 083.
+- [x] T004 Record requirements/product-vision drift review.
+
+## Phase 2: Generic Acquisition Guidance
+
+- [x] T005 Add focused failing test for tool acquisition guidance.
+- [x] T006 Add generic acquisition/risk fields to `oss-plan.json` tool plans.
+- [x] T007 Update answer/query guidance to reject stack-owned adapters while
+  still recommending local producer tools.
+
+## Phase 3: Stress And Verification
+
+- [x] T008 Run focused `go test ./internal/contextprep`.
+- [x] T009 Run full baseline.
+- [x] T010 Run fresh Bigtop context smoke.
+- [x] T011 Run Cursor Composer 2.5 bounded stress.
+- [x] T012 Run independent review lanes and record disposition.
+- [x] T013 Commit, push, create PR, and refresh check state.
diff --git a/internal/contextprep/contextprep.go b/internal/contextprep/contextprep.go
index 6dc2ddd..5999dff 100644
--- a/internal/contextprep/contextprep.go
+++ b/internal/contextprep/contextprep.go
@@ -188,16 +188,26 @@ type ossPlanFile struct {
 }
 
 type OSSToolPlan struct {
-	ID                string       `json:"id"`
-	Family            string       `json:"family"`
-	Producer          string       `json:"producer"`
-	Purpose           string       `json:"purpose"`
-	Executable        string       `json:"executable"`
-	AgentCapabilities []string     `json:"agent_capabilities,omitempty"`
-	Status            string       `json:"status"`
-	EvidenceState     string       `json:"evidence_state"`
-	Reason            string       `json:"reason"`
-	Commands          []OSSCommand `json:"commands,omitempty"`
+	ID                string          `json:"id"`
+	Family            string          `json:"family"`
+	Producer          string          `json:"producer"`
+	Purpose           string          `json:"purpose"`
+	Executable        string          `json:"executable"`
+	AgentCapabilities []string        `json:"agent_capabilities,omitempty"`
+	Status            string          `json:"status"`
+	EvidenceState     string          `json:"evidence_state"`
+	Reason            string          `json:"reason"`
+	Acquisition       ToolAcquisition `json:"acquisition"`
+	Commands          []OSSCommand    `json:"commands,omitempty"`
+}
+
+type ToolAcquisition struct {
+	Kind                string   `json:"kind"`
+	Availability        string   `json:"availability"`
+	NextAction          string   `json:"next_action"`
+	EvidenceUntilOutput string   `json:"evidence_until_output"`
+	Risks               []string `json:"risks"`
+	Boundary            string   `json:"boundary"`
 }
 
 type OSSCommand struct {
@@ -1778,6 +1788,9 @@ func buildOSSPlan(root, out, profile string, tools []ToolEntry, buildTools build
 	sort.Slice(plan.Tools, func(i, j int) bool {
 		return plan.Tools[i].ID < plan.Tools[j].ID
 	})
+	for i := range plan.Tools {
+		plan.Tools[i].Acquisition = toolAcquisition(plan.Tools[i])
+	}
 	return plan
 }
 
@@ -1789,6 +1802,88 @@ func toolFamiliesPresent(tools []ToolEntry) map[string]bool {
 	return present
 }
 
+func toolAcquisition(tool OSSToolPlan) ToolAcquisition {
+	acquisition := ToolAcquisition{
+		Kind:                "native-producer-tool",
+		Availability:        acquisitionAvailability(tool),
+		NextAction:          acquisitionNextAction(tool),
+		EvidenceUntilOutput: "not_assessed",
+		Risks:               acquisitionRisks(tool),
+		Boundary:            "candidate tool guidance is stack-agnostic and does not create a Portolan-owned language adapter; local output must be produced and re-ingested before claims are upgraded",
+	}
+	if tool.Status == "input_present" {
+		acquisition.EvidenceUntilOutput = tool.EvidenceState
+		acquisition.NextAction = "inspect existing local output before acquiring or running another producer"
+	}
+	return acquisition
+}
+
+func acquisitionAvailability(tool OSSToolPlan) string {
+	switch tool.Status {
+	case "available_not_run":
+		return "installed"
+	case "not_assessed":
+		if tool.Executable != "" {
+			return "installed_but_requires_local_evaluation"
+		}
+		return "requires_local_evaluation"
+	case "not_available":
+		return "missing"
+	case "input_present":
+		return "local_output_present"
+	default:
+		return "unknown"
+	}
+}
+
+func acquisitionNextAction(tool OSSToolPlan) string {
+	switch tool.Status {
+	case "available_not_run":
+		if len(tool.Commands) > 0 {
+			return "request approval, run the bounded native producer command, then rerun context prepare --force"
+		}
+		return "request approval and evaluate the local producer before running it"
+	case "not_available":
+		return "install or expose the local producer tool only after approval, then rerun context prepare --force"
+	case "not_assessed":
+		return "evaluate the local producer configuration and output path boundary before suggesting a command"
+	case "input_present":
+		return "inspect existing local output before acquiring or running another producer"
+	default:
+		return "preserve not_assessed and ask for local producer output"
+	}
+}
+
+func acquisitionRisks(tool OSSToolPlan) []string {
+	risks := []string{
+		"candidate tool output is not evidence until local output exists and is re-ingested",
+		"do not install, fetch, or update tools without explicit approval",
+	}
+	if tool.Executable == "" && tool.Status == "not_available" {
+		risks = append(risks, "tool is missing from PATH")
+	}
+	if len(tool.Commands) == 0 {
+		risks = append(risks, "no bounded command is currently declared")
+	}
+	network := ""
+	mutatesTarget := false
+	for _, command := range tool.Commands {
+		if command.Network != "" && command.Network != "not_expected" && command.Network != "not_expected_for_local_filesystem_source" {
+			network = command.Network
+		}
+		if command.MutatesTarget {
+			mutatesTarget = true
+		}
+	}
+	if network != "" {
+		risks = append(risks, "network may be used: "+network)
+	}
+	if mutatesTarget {
+		risks = append(risks, "native command may mutate target files or tool caches")
+	}
+	return risks
+}
+
 func jscpdPlan(root, out, toolOutputDir string, inputPresent bool, repos []Repository) OSSToolPlan {
 	plan := baseOSSPlan("jscpd", "jscpd", "jscpd", "Detect duplicated source and text fragments as metadata-visible duplication evidence.")
 	plan.AgentCapabilities = []string{
@@ -2336,7 +2431,8 @@ func renderAnswerContract(root, out string) string {
 	fmt.Fprintf(&b, "Relationship answers must name both relationship kind and evidence type. For relationship claims, including \"what talks to what?\", look first at `evidence-index.jsonl`, `tool-registry.json`, `gaps.jsonl`, then map-bundle `summary.json`, `graph-index.json`, and `findings.jsonl`. `evidence-index.jsonl` may include build/deploy relationship candidates such as build manifests, distribution manifests, RPM specs, and deployment manifests; those are source-visible places to inspect, not parsed service topology. Native map relationship extraction is limited to Go imports and go.mod manifests; JVM, PHP, Scala, and other non-Go coupling stays `not_assessed` unless supplied through local producer output. `source-visible` and `metadata-visible` records do not prove runtime communication; runtime topology is `not_assessed` unless runtime-visible local observations were supplied and inspected. Dependency and symbol records from local producer outputs do not mean Portolan has native PHP, JVM, Scala, or other language semantics; they are producer evidence. Missing relationship surfaces remain `unknown`, `cannot_verify`, or `not_assessed`; `claim-only` remains a claim, not observed evidence.\n\n")
 	fmt.Fprintf(&b, "## Producer Family Recommendations\n\n")
 	fmt.Fprintf(&b, "Producer recommendations are options, not observed evidence. Treat `producer-recommendation` records in `evidence-index.jsonl` as a safe next-action surface for missing local producer families; they do not prove the candidate tool is installed, supported, or appropriate for this landscape. Candidate tools marked `candidate_only` remain `not_assessed` until local output or a local evaluation record exists. Portolan does not synthesize producer evaluations from recommendation records; if no `producer-evaluation` record is present, candidate evaluation remains `not_assessed`. Check both `verification_state` and `support_state`: `verification_state` describes local evidence for the candidate, while `support_state` describes whether Portolan can present it as supported. Do not propose a Portolan-owned PHP/JVM/Scala adapter as the default answer to language coverage gaps; ask for or evaluate local dependency, symbol-index, API/catalog, deployment/model, static finding, duplication, config, or runtime-observation producer evidence instead.\n\n")
-	fmt.Fprintf(&b, "Native Maven/Gradle build-tool producer output is the preferred first step for visible JVM build manifests. Java/Scala/Maven dependency relationships remain `not_assessed` until local producer output exists and is inspected. `oss-plan.json` may list Maven or Gradle CycloneDX producer options when `pom.xml`, `build.gradle`, or `build.gradle.kts` files are visible, but those options are approval-required and do not imply Portolan executed Maven, Gradle, or a `portolan produce` command. Do not turn a visible `pom.xml` or `build.gradle` into a Portolan-owned JVM adapter request; keep it as a native producer-evidence acquisition question.\n\n")
+	fmt.Fprintf(&b, "Tool acquisition guidance is stack-agnostic. Candidate tools are local producer options, not Portolan-owned language adapters. Use `oss-plan.json` acquisition fields to decide whether a tool is installed, missing, or requires local evaluation, and to name approval, network, cache, mutation, and output-path risks before asking the operator to pull it in. Do not propose a Portolan-owned PHP/JVM/Scala/Gradle adapter as the default answer to language coverage gaps; ask for local producer output from an appropriate existing tool family and keep evidence `not_assessed` until that output is present and re-ingested.\n\n")
+	fmt.Fprintf(&b, "Build-system and SBOM producer output can be the right first evidence path when dependency/build manifests are visible, but the choice is landscape-specific and approval-gated. Java, Scala, Maven, Gradle, PHP, Composer, and other ecosystem relationships remain `not_assessed` until local producer output exists and is inspected. `oss-plan.json` may list concrete options such as Maven, Gradle, Composer, Syft, or CycloneDX-compatible producers when matching local surfaces are visible, but those options do not imply Portolan executed the tool or owns that stack. Keep visible manifests as native producer-evidence acquisition questions, not requests for Portolan-owned language or build-system adapters.\n\n")
 	fmt.Fprintf(&b, "## Producer Run Records\n\n")
 	fmt.Fprintf(&b, "`producer-run` records in `evidence-index.jsonl` describe externally generated local outputs selected by the operator. They are not Portolan execution receipts and they do not imply a `portolan produce` command exists. A `verified` producer-run record proves only that the referenced local output file existed during context preparation and that the record passed Portolan's metadata validation. Use `producer_family`, `producer_tool`, `output_path`, `scope`, `covered_units`, `freshness`, and `limitations` before making a claim. Static `deployment-model` and `api-catalog` records stay `metadata-visible`; they must not be promoted to `runtime-visible` or to whole-landscape coverage. Runtime topology stays `not_assessed` unless a runtime producer family supplies `runtime-visible` local observations.\n\n")
 	fmt.Fprintf(&b, "## Hard Boundaries\n\n")
@@ -2373,17 +2469,21 @@ func renderQueryPlan() string {
   records with kind ` + "`relationship-candidate`" + ` before opening raw source.
   They point at build manifests, distribution manifests, RPM specs, and
   deployment manifests; they are candidate evidence, not parsed topology.
-- Maven/Gradle dependency evidence: if build manifests are visible but
+- Dependency producer evidence: if dependency/build manifests are visible but
   CycloneDX/build-tool outputs are missing, inspect ` + "`oss-plan.json`" + ` for
-  ` + "`maven-cyclonedx`" + ` and ` + "`gradle-cyclonedx`" + ` plans. Treat them as
-  approval-gated native producer options, not Portolan-owned JVM adapters, and
-  keep dependency relationships ` + "`not_assessed`" + ` until local output exists.
+  build-system, SBOM, or CycloneDX-compatible plans. Treat them as
+  approval-gated native producer options, not Portolan-owned stack adapters,
+  and keep dependency relationships ` + "`not_assessed`" + ` until local output exists.
 - Producer family gaps: inspect ` + "`producer-coverage`" + ` and
   ` + "`producer-recommendation`" + ` records in ` + "`evidence-index.jsonl`" + ` before
   making mixed-language coverage claims. Recommendations are options, not
   verified support. If no ` + "`producer-evaluation`" + ` records are present,
   candidate evaluation remains not_assessed; Portolan does not synthesize
   evaluations from recommendations.
+- Tool acquisition: inspect ` + "`oss-plan.json`" + ` acquisition fields before
+  asking the operator to pull in a local tool. Treat the tool as a candidate
+  producer, not as a Portolan-owned stack adapter. Name availability, approval,
+  network/cache/mutation risks, and the required post-run context refresh.
 - Implicit knowledge: inspect repository manifests, local catalogs, contracts,
   and index handles. Do not turn naming conventions into facts without evidence.
 - Service relationships: start with Backstage, OpenAPI, AsyncAPI, Structurizr,
diff --git a/internal/contextprep/contextprep_test.go b/internal/contextprep/contextprep_test.go
index 84dfd24..91497cf 100644
--- a/internal/contextprep/contextprep_test.go
+++ b/internal/contextprep/contextprep_test.go
@@ -104,17 +104,89 @@ func TestRunAddsBuildToolDependencyProducerPlans(t *testing.T) {
 
 	contract := mustReadContextprep(t, filepath.Join(out, "answer-contract.md"))
 	for _, want := range []string{
-		"Native Maven/Gradle build-tool producer output",
-		"Java/Scala/Maven dependency relationships remain `not_assessed`",
-		"Do not turn a visible `pom.xml` or `build.gradle` into a Portolan-owned JVM adapter request",
+		"Build-system and SBOM producer output can be the right first evidence path",
+		"Java, Scala, Maven, Gradle, PHP, Composer, and other ecosystem relationships remain `not_assessed`",
+		"not requests for Portolan-owned language or build-system adapters",
 	} {
 		if !strings.Contains(contract, want) {
 			t.Fatalf("answer-contract.md missing %q:\n%s", want, contract)
 		}
 	}
 	queryPlan := mustReadContextprep(t, filepath.Join(out, "query-plan.md"))
-	if !strings.Contains(queryPlan, "Maven/Gradle dependency evidence") {
-		t.Fatalf("query-plan.md missing Maven/Gradle dependency guidance:\n%s", queryPlan)
+	if !strings.Contains(queryPlan, "Dependency producer evidence") {
+		t.Fatalf("query-plan.md missing dependency producer guidance:\n%s", queryPlan)
+	}
+}
+
+func TestRunWritesStackAgnosticToolAcquisitionGuidance(t *testing.T) {
+	root := t.TempDir()
+	mavenRepo := filepath.Join(root, "repos", "maven-service")
+	gradleRepo := filepath.Join(root, "repos", "gradle-service")
+	mustMkdirContextprep(t, filepath.Join(mavenRepo, ".git"))
+	mustMkdirContextprep(t, filepath.Join(gradleRepo, ".git"))
+	mustWriteContextprep(t, filepath.Join(mavenRepo, "pom.xml"), `<project><artifactId>maven-service</artifactId></project>`)
+	mustWriteContextprep(t, filepath.Join(gradleRepo, "build.gradle.kts"), `plugins { java }`)
+
+	bin := filepath.Join(t.TempDir(), "bin")
+	mustMkdirContextprep(t, bin)
+	for _, name := range []string{"mvn", "gradle", "syft", "jscpd"} {
+		path := filepath.Join(bin, name)
+		mustWriteContextprep(t, path, "#!/bin/sh\nexit 0\n")
+		if err := os.Chmod(path, 0o755); err != nil {
+			t.Fatal(err)
+		}
+	}
+	t.Setenv("PATH", bin)
+
+	out := filepath.Join(root, ".portolan", "context")
+	if _, err := Run(Options{RootPath: root, OutputPath: out, Profile: "agent"}); err != nil {
+		t.Fatalf("Run returned error: %v", err)
+	}
+
+	plan := readOSSPlanContextprep(t, filepath.Join(out, "oss-plan.json"))
+	if len(plan.Tools) == 0 {
+		t.Fatal("oss-plan has no tool candidates")
+	}
+	for _, tool := range plan.Tools {
+		if tool.Acquisition.Kind == "" {
+			t.Fatalf("tool %s missing acquisition guidance: %#v", tool.ID, tool)
+		}
+		if tool.Acquisition.EvidenceUntilOutput != "not_assessed" {
+			t.Fatalf("tool %s acquisition evidence = %q, want not_assessed", tool.ID, tool.Acquisition.EvidenceUntilOutput)
+		}
+		if len(tool.Acquisition.Risks) == 0 {
+			t.Fatalf("tool %s missing acquisition risks: %#v", tool.ID, tool.Acquisition)
+		}
+	}
+
+	maven := toolPlanByIDContextprep(plan, "maven-cyclonedx")
+	if maven.Acquisition.Kind != "native-producer-tool" {
+		t.Fatalf("maven acquisition = %#v, want native producer tool", maven.Acquisition)
+	}
+	if !strings.Contains(maven.Acquisition.NextAction, "approval") {
+		t.Fatalf("maven next action = %q, want approval boundary", maven.Acquisition.NextAction)
+	}
+	gradle := toolPlanByIDContextprep(plan, "gradle-cyclonedx")
+	if gradle.Acquisition.Kind != "native-producer-tool" {
+		t.Fatalf("gradle acquisition = %#v, want native producer tool", gradle.Acquisition)
+	}
+	if !strings.Contains(gradle.Acquisition.NextAction, "evaluate") {
+		t.Fatalf("gradle next action = %q, want local evaluation boundary", gradle.Acquisition.NextAction)
+	}
+
+	contract := mustReadContextprep(t, filepath.Join(out, "answer-contract.md"))
+	for _, want := range []string{
+		"Tool acquisition guidance is stack-agnostic",
+		"Candidate tools are local producer options, not Portolan-owned language adapters",
+		"Do not propose a Portolan-owned PHP/JVM/Scala/Gradle adapter",
+	} {
+		if !strings.Contains(contract, want) {
+			t.Fatalf("answer-contract.md missing %q:\n%s", want, contract)
+		}
+	}
+	queryPlan := mustReadContextprep(t, filepath.Join(out, "query-plan.md"))
+	if !strings.Contains(queryPlan, "Tool acquisition") {
+		t.Fatalf("query-plan.md missing tool acquisition guidance:\n%s", queryPlan)
 	}
 }
 
@@ -141,6 +213,15 @@ func TestResolveBuildToolExecutablePrefersExecutableWrapper(t *testing.T) {
 	}
 }
 
+func toolPlanByIDContextprep(plan ossPlanFile, id string) OSSToolPlan {
+	for _, tool := range plan.Tools {
+		if tool.ID == id {
+			return tool
+		}
+	}
+	return OSSToolPlan{}
+}
+
 func TestMavenCycloneDXPlanCapsRepositoryCommands(t *testing.T) {
 	root := t.TempDir()
 	bin := filepath.Join(t.TempDir(), "bin")