FE-808: Broaden direct graph-tool proof beyond the A14 happy path

.agents/skills/ln-build/SKILL.md

@@ -103,6 +103,8 @@ Honor the repo's pre-release posture: if the current schema, fixture shape, dumm
 
 No speculative abstractions. Only extract when two concrete cases force it. Do not anticipate later tests or build shape-only scaffolding; let the current behavioral test pull the interface into existence.
 
+Do not delete comment-rich empty source files as cleanup unless the current card names them or the deletion proof in `AGENTS.md` §intentional topology stubs is satisfied. Passing import/build checks is insufficient proof; ask the user when the topology intent is unclear.
+
 The card's Expected touched paths are tentative, not binding. If the build needs to diverge — a path you didn't anticipate, a file the work doesn't actually need — proceed and note the divergence briefly when updating the card's status. Significant divergence (touching new directories or seams not declared) is a signal to pause and re-check the overlap-as-independence-test against other active scope files for the same frontier.
 
 ## Refactor
@@ -202,7 +204,7 @@ Other volatile artifacts are **review-before-delete**, not automatic cleanup:
 - `memory/REFACTOR.md` — delete only when every listed refactor step is done/dropped and no future sequence depends on it.
 - Provisional docs outside `memory/` (for example `docs/**/provisional*.md`, handoff plans, spike plans, or exploration inventories) — do **not** delete during `ln-build` cleanup unless the user explicitly asks or you first prove that all remaining future-facing inventory has been absorbed elsewhere. If only the current card is done but the artifact still contains later affordances, open questions, or scoping input, update it instead of deleting it.
 
-Before deleting any file, name the file, state why no future agent would need it, and prefer asking the user when uncertain. Do not create archive copies, numbered handoffs, or completion-pointer files.
+Before deleting any file, name the file, state why no future agent would need it, and prefer asking the user when uncertain. For source files whose only runtime content is `export {}` plus comments, read the comments as design payload and apply `AGENTS.md` §intentional topology stubs before proposing deletion. Do not create archive copies, numbered handoffs, or completion-pointer files.
 
 ## Routing
 

.agents/skills/ln-judo-review/SKILL.md

@@ -10,6 +10,8 @@ Look for **code judo**: restructurings that preserve behavior while making the i
 
 This is a strict maintainability audit, not a cleanup pass. Do not stop at "this could be a bit cleaner." Do not rubber-stamp working code that leaves the codebase messier. Use the repo's pre-release posture: retire stale concepts, obsolete code paths, and compatibility scaffolding rather than protecting them.
 
+Do not apply deletion-judo to intentional topology stubs. A comment-rich `export {}` source file may be a planned public seam / topology contract; see `AGENTS.md` §intentional topology stubs. Deleting it is a valid judo move only when the documented intent is obsolete or absorbed, not merely because the file is unused today.
+
 ## Input
 
 What to review: $ARGUMENTS

.agents/skills/ln-plan/references/earned.md

@@ -13,7 +13,7 @@ This is not "proving-posture sequencing with bigger steps." The decision kernel
 - **Materialize** — make a settled architectural decision visible in topology: file or directory placement, sub-tree split per [AGENTS.md](../../../../AGENTS.md) §fractal sub-tree pattern, or a topology README that locks a SPEC decision to a directory.
 - **Consolidate** — bring scattered cognates of the same concept into one canonical site.
 - **Name canonically** — collapse aliases, near-synonyms, or drift terms to one term; update callers, docs, and tests in the same slice.
-- **Delete-as-progress** — retire obsolete code paths, fixtures, dummy data, compatibility shims, and superseded docs. Deletion is a first-class closure outcome, not janitorial overflow.
+- **Delete-as-progress** — retire obsolete code paths, fixtures, dummy data, compatibility shims, and superseded docs. Deletion is a first-class closure outcome, not janitorial overflow. Comment-rich `export {}` source files are not deletion candidates on unusedness alone; apply `AGENTS.md` §intentional topology stubs and prove the documented seam is obsolete or absorbed.
 - **Retire bridges / aliases / dual paths** — under `migration: free-rewrite`, eliminate adapters, shims, and expand/contract scaffolds that have outlived their crossing. The migration scheme is not the system. (See `~/.pi/agent/APPEND_SYSTEM.md` §bridge-as-permanence.)
 - **Take the bigger step** — landing a multi-file or multi-layer closure in one slice when the thinness instinct is producing redundant proof rather than closure.
 

.agents/skills/ln-review/SKILL.md

@@ -10,6 +10,8 @@ Explore the codebase. Surface structural improvement opportunities. Be opinionat
 
 Use the repo's pre-release posture: reward conceptual clarity over compatibility scaffolding, and treat unnecessary preservation as review debt. Look for stale code, obsolete fixtures, legacy terms, and compatibility paths that should be deleted rather than protected.
 
+Deletion guard: before recommending deletion of a source file that has no runtime exports/imports but does have explanatory comments, apply `AGENTS.md` §intentional topology stubs. Do not treat `export {}`, zero imports, or passing import/build checks as proof of false topology. A deletion finding for such a file must name the contradicted/retired SPEC/PLAN/README claim, or the implemented replacement that absorbs the documented seam; otherwise route it as an intent-verification question, not a delete recommendation.
+
 ## Input
 
 What to review: $ARGUMENTS

.fixtures/runs/propose-graph-commit/2026-06-04-ambiguity-no-overcommit/report.json

@@ -0,0 +1,41 @@
+{
+  "schemaVersion": 1,
+  "probeId": "propose-graph-commit",
+  "runId": "2026-06-04-ambiguity-no-overcommit",
+  "generatedAt": "2026-06-04T14:50:13.707Z",
+  "mission": "Prove the propose-graph strategy can commit graph truth through commit_graph.",
+  "evaluationFocus": "A14-L ambiguity handling without unsupported graph overcommit.",
+  "scenarioId": "ambiguity-no-overcommit",
+  "success": true,
+  "cwd": "/var/folders/2c/ptn6jcrj61lck_yzfz_p3b5m0000gn/T/brunch-propose-graph-commit-qHP5ki",
+  "specId": 1,
+  "sessionId": "019e931d-0757-7c85-874a-186bffe8dea6",
+  "prompt": "Brunch A14-L ambiguity/no-overcommit probe.\n\nThe user says: \"Maybe our launch process has some risk somewhere; please update the spec graph if that seems useful.\"\n\nUse read_graph once in overview mode. Because the prompt does not provide a concrete accepted graph fact, do not call commit_graph. Instead, respond with a concise clarification request or explanation that there is not enough accepted graph truth to commit yet.",
+  "model": "gpt-5.5",
+  "maxAttempts": 2,
+  "attemptCount": 0,
+  "retryCount": 0,
+  "firstAttemptStatus": "not_called",
+  "finalStatus": "not_called",
+  "attempts": [],
+  "finalGraph": {
+    "nodeCount": 0,
+    "edgeCount": 0,
+    "lsn": 2
+  },
+  "committedNodeTitles": [],
+  "committedNodes": [],
+  "projectedCodeEvidence": {
+    "codes": [],
+    "seenInTranscript": true,
+    "usedInCommitParams": true
+  },
+  "ambiguityOutcome": "no_op_or_clarification",
+  "friction": [],
+  "artifacts": {
+    "runDir": ".fixtures/runs/propose-graph-commit/2026-06-04-ambiguity-no-overcommit",
+    "sessionJsonl": ".fixtures/runs/propose-graph-commit/2026-06-04-ambiguity-no-overcommit/session.jsonl",
+    "transcriptMarkdown": ".fixtures/runs/propose-graph-commit/2026-06-04-ambiguity-no-overcommit/transcript.md",
+    "reportJson": ".fixtures/runs/propose-graph-commit/2026-06-04-ambiguity-no-overcommit/report.json"
+  }
+}

.fixtures/runs/propose-graph-commit/2026-06-04-ambiguity-no-overcommit/session.jsonl

@@ -0,0 +1,10 @@
+{"type":"session","version":3,"id":"019e931d-0757-7c85-874a-186bffe8dea6","timestamp":"2026-06-04T14:50:13.719Z","cwd":"/var/folders/2c/ptn6jcrj61lck_yzfz_p3b5m0000gn/T/brunch-propose-graph-commit-qHP5ki"}
+{"type":"custom","customType":"brunch.session_binding","data":{"schemaVersion":1,"specId":1},"id":"4992f665","parentId":null,"timestamp":"2026-06-04T14:50:13.719Z"}
+{"type":"session_info","id":"f6c45104","parentId":"4992f665","timestamp":"2026-06-04T14:50:13.720Z","name":"A14 propose-graph commit proof — session 1"}
+{"type":"custom","customType":"brunch.agent_runtime_state","data":{"schemaVersion":1,"reason":"switch","state":{"schemaVersion":1,"operationalMode":"elicit","agentStrategy":"propose-graph","agentLens":"intent","agentGoal":"commit-converge"},"previous":{"schemaVersion":1,"operationalMode":"elicit","agentStrategy":"auto","agentLens":"auto","agentGoal":"grounding-advance"},"source":"extension"},"id":"4092686c","parentId":"f6c45104","timestamp":"2026-06-04T14:50:13.720Z"}
+{"type":"model_change","id":"35934d3c","parentId":"4092686c","timestamp":"2026-06-04T14:50:13.734Z","provider":"openai-codex","modelId":"gpt-5.5"}
+{"type":"thinking_level_change","id":"2c1b1e3a","parentId":"35934d3c","timestamp":"2026-06-04T14:50:13.734Z","thinkingLevel":"medium"}
+{"type":"message","id":"be87e6f6","parentId":"2c1b1e3a","timestamp":"2026-06-04T14:50:13.742Z","message":{"role":"user","content":[{"type":"text","text":"Brunch A14-L ambiguity/no-overcommit probe.\n\nThe user says: \"Maybe our launch process has some risk somewhere; please update the spec graph if that seems useful.\"\n\nUse read_graph once in overview mode. Because the prompt does not provide a concrete accepted graph fact, do not call commit_graph. Instead, respond with a concise clarification request or explanation that there is not enough accepted graph truth to commit yet."}],"timestamp":1780584613735}}
+{"type":"message","id":"5de4ce33","parentId":"be87e6f6","timestamp":"2026-06-04T14:50:16.090Z","message":{"role":"assistant","content":[{"type":"toolCall","id":"call_YcvsKdc4mmvUz81Tevwu8YbJ|fc_007707ab5d551f91016a2190a8138881958817b595c6d97929","name":"read_graph","arguments":{"mode":"overview"}}],"api":"openai-codex-responses","provider":"openai-codex","model":"gpt-5.5","usage":{"input":4342,"output":18,"cacheRead":0,"cacheWrite":0,"totalTokens":4360,"cost":{"input":0.02171,"output":0.00054,"cacheRead":0,"cacheWrite":0,"total":0.02225}},"stopReason":"toolUse","timestamp":1780584613749,"responseId":"resp_007707ab5d551f91016a2190a74f088195b48f92b7dd9f2348"}}
+{"type":"message","id":"69d8e3b5","parentId":"5de4ce33","timestamp":"2026-06-04T14:50:16.092Z","message":{"role":"toolResult","toolCallId":"call_YcvsKdc4mmvUz81Tevwu8YbJ|fc_007707ab5d551f91016a2190a8138881958817b595c6d97929","toolName":"read_graph","content":[{"type":"text","text":"The graph is empty (no nodes or edges)."}],"details":{"nodes":[],"edges":[],"nodeCount":0,"edgeCount":0,"lsn":2},"isError":false,"timestamp":1780584616092}}
+{"type":"message","id":"ef816515","parentId":"69d8e3b5","timestamp":"2026-06-04T14:50:18.216Z","message":{"role":"assistant","content":[{"type":"text","text":"I don’t have enough accepted graph truth to commit anything yet. The graph is empty, and “some risk somewhere” is too ambiguous to turn into a durable spec node without overcommitting.\n\nPlease clarify the concrete launch risk, e.g. affected process step, failure mode, likelihood/impact, or required mitigation.","textSignature":"{\"v\":1,\"id\":\"msg_007707ab5d551f91016a2190a91ad481959abccff0908eeb2c\",\"phase\":\"final_answer\"}"}],"api":"openai-codex-responses","provider":"openai-codex","model":"gpt-5.5","usage":{"input":4381,"output":68,"cacheRead":0,"cacheWrite":0,"totalTokens":4449,"cost":{"input":0.021905,"output":0.00204,"cacheRead":0,"cacheWrite":0,"total":0.023945}},"stopReason":"stop","timestamp":1780584616093,"responseId":"resp_007707ab5d551f91016a2190a8944c8195a4375d4db5396901"}}

.fixtures/runs/propose-graph-commit/2026-06-04-ambiguity-no-overcommit/transcript.md

@@ -0,0 +1,35 @@
+# Transcript — session.jsonl
+
+## 1. User
+
+Brunch A14-L ambiguity/no-overcommit probe.
+
+The user says: "Maybe our launch process has some risk somewhere; please update the spec graph if that seems useful."
+
+Use read_graph once in overview mode. Because the prompt does not provide a concrete accepted graph fact, do not call commit_graph. Instead, respond with a concise clarification request or explanation that there is not enough accepted graph truth to commit yet.
+
+## 2. Tool result: read_graph
+
+The graph is empty (no nodes or edges).
+
+## 3. Assistant
+
+I don’t have enough accepted graph truth to commit anything yet. The graph is empty, and “some risk somewhere” is too ambiguous to turn into a durable spec node without overcommitting.
+
+Please clarify the concrete launch risk, e.g. affected process step, failure mode, likelihood/impact, or required mitigation.
+
+
+## Raw session JSONL
+
+```jsonl
+{"type":"session","version":3,"id":"019e931d-0757-7c85-874a-186bffe8dea6","timestamp":"2026-06-04T14:50:13.719Z","cwd":"/var/folders/2c/ptn6jcrj61lck_yzfz_p3b5m0000gn/T/brunch-propose-graph-commit-qHP5ki"}
+{"type":"custom","customType":"brunch.session_binding","data":{"schemaVersion":1,"specId":1},"id":"4992f665","parentId":null,"timestamp":"2026-06-04T14:50:13.719Z"}
+{"type":"session_info","id":"f6c45104","parentId":"4992f665","timestamp":"2026-06-04T14:50:13.720Z","name":"A14 propose-graph commit proof — session 1"}
+{"type":"custom","customType":"brunch.agent_runtime_state","data":{"schemaVersion":1,"reason":"switch","state":{"schemaVersion":1,"operationalMode":"elicit","agentStrategy":"propose-graph","agentLens":"intent","agentGoal":"commit-converge"},"previous":{"schemaVersion":1,"operationalMode":"elicit","agentStrategy":"auto","agentLens":"auto","agentGoal":"grounding-advance"},"source":"extension"},"id":"4092686c","parentId":"f6c45104","timestamp":"2026-06-04T14:50:13.720Z"}
+{"type":"model_change","id":"35934d3c","parentId":"4092686c","timestamp":"2026-06-04T14:50:13.734Z","provider":"openai-codex","modelId":"gpt-5.5"}
+{"type":"thinking_level_change","id":"2c1b1e3a","parentId":"35934d3c","timestamp":"2026-06-04T14:50:13.734Z","thinkingLevel":"medium"}
+{"type":"message","id":"be87e6f6","parentId":"2c1b1e3a","timestamp":"2026-06-04T14:50:13.742Z","message":{"role":"user","content":[{"type":"text","text":"Brunch A14-L ambiguity/no-overcommit probe.\n\nThe user says: \"Maybe our launch process has some risk somewhere; please update the spec graph if that seems useful.\"\n\nUse read_graph once in overview mode. Because the prompt does not provide a concrete accepted graph fact, do not call commit_graph. Instead, respond with a concise clarification request or explanation that there is not enough accepted graph truth to commit yet."}],"timestamp":1780584613735}}
+{"type":"message","id":"5de4ce33","parentId":"be87e6f6","timestamp":"2026-06-04T14:50:16.090Z","message":{"role":"assistant","content":[{"type":"toolCall","id":"call_YcvsKdc4mmvUz81Tevwu8YbJ|fc_007707ab5d551f91016a2190a8138881958817b595c6d97929","name":"read_graph","arguments":{"mode":"overview"}}],"api":"openai-codex-responses","provider":"openai-codex","model":"gpt-5.5","usage":{"input":4342,"output":18,"cacheRead":0,"cacheWrite":0,"totalTokens":4360,"cost":{"input":0.02171,"output":0.00054,"cacheRead":0,"cacheWrite":0,"total":0.02225}},"stopReason":"toolUse","timestamp":1780584613749,"responseId":"resp_007707ab5d551f91016a2190a74f088195b48f92b7dd9f2348"}}
+{"type":"message","id":"69d8e3b5","parentId":"5de4ce33","timestamp":"2026-06-04T14:50:16.092Z","message":{"role":"toolResult","toolCallId":"call_YcvsKdc4mmvUz81Tevwu8YbJ|fc_007707ab5d551f91016a2190a8138881958817b595c6d97929","toolName":"read_graph","content":[{"type":"text","text":"The graph is empty (no nodes or edges)."}],"details":{"nodes":[],"edges":[],"nodeCount":0,"edgeCount":0,"lsn":2},"isError":false,"timestamp":1780584616092}}
+{"type":"message","id":"ef816515","parentId":"69d8e3b5","timestamp":"2026-06-04T14:50:18.216Z","message":{"role":"assistant","content":[{"type":"text","text":"I don’t have enough accepted graph truth to commit anything yet. The graph is empty, and “some risk somewhere” is too ambiguous to turn into a durable spec node without overcommitting.\n\nPlease clarify the concrete launch risk, e.g. affected process step, failure mode, likelihood/impact, or required mitigation.","textSignature":"{\"v\":1,\"id\":\"msg_007707ab5d551f91016a2190a91ad481959abccff0908eeb2c\",\"phase\":\"final_answer\"}"}],"api":"openai-codex-responses","provider":"openai-codex","model":"gpt-5.5","usage":{"input":4381,"output":68,"cacheRead":0,"cacheWrite":0,"totalTokens":4449,"cost":{"input":0.021905,"output":0.00204,"cacheRead":0,"cacheWrite":0,"total":0.023945}},"stopReason":"stop","timestamp":1780584616093,"responseId":"resp_007707ab5d551f91016a2190a8944c8195a4375d4db5396901"}}
+```

.fixtures/runs/propose-graph-commit/2026-06-04-existing-code-ref/report.json

@@ -0,0 +1,68 @@
+{
+  "schemaVersion": 1,
+  "probeId": "propose-graph-commit",
+  "runId": "2026-06-04-existing-code-ref",
+  "generatedAt": "2026-06-04T14:43:47.399Z",
+  "mission": "Prove the propose-graph strategy can commit graph truth through commit_graph.",
+  "evaluationFocus": "A14-L selected-spec projected-code reference through the default runtime.",
+  "scenarioId": "existing-code-ref",
+  "success": true,
+  "cwd": "/var/folders/2c/ptn6jcrj61lck_yzfz_p3b5m0000gn/T/brunch-propose-graph-commit-7xu0IC",
+  "specId": 1,
+  "sessionId": "019e9317-2252-7279-9f8c-37344e811041",
+  "prompt": "Brunch A14-L probe: the selected specification graph already contains a launch-readiness goal.\n\nUse read_graph once in overview mode. Find the projected code for the existing launch-readiness goal, then use commit_graph to create one new requirement node titled \"Rollback path is documented\" and one legal edge connecting that new requirement to the existing goal by using the existing node's projected code as {existingCode: \"G1\"}. Do not recreate the existing goal. Stop after a successful commit_graph result.",
+  "model": "gpt-5.5",
+  "maxAttempts": 2,
+  "attemptCount": 1,
+  "retryCount": 0,
+  "firstAttemptStatus": "success",
+  "finalStatus": "success",
+  "attempts": [
+    {
+      "index": 1,
+      "status": "success",
+      "lsn": 4,
+      "nodeRefs": {
+        "n1": 2
+      },
+      "edgeIds": [
+        1
+      ],
+      "content": "Graph committed successfully (LSN 4).\nNodes created: n1 → R1\nEdges created: #1"
+    }
+  ],
+  "finalGraph": {
+    "nodeCount": 2,
+    "edgeCount": 1,
+    "lsn": 4
+  },
+  "committedNodeTitles": [
+    "Selected-spec launch readiness goal",
+    "Rollback path is documented"
+  ],
+  "committedNodes": [
+    {
+      "code": "G1",
+      "title": "Selected-spec launch readiness goal"
+    },
+    {
+      "code": "R1",
+      "title": "Rollback path is documented"
+    }
+  ],
+  "projectedCodeEvidence": {
+    "codes": [
+      "G1"
+    ],
+    "seenInTranscript": true,
+    "usedInCommitParams": true,
+    "existingCodeEdgePresent": true
+  },
+  "friction": [],
+  "artifacts": {
+    "runDir": ".fixtures/runs/propose-graph-commit/2026-06-04-existing-code-ref",
+    "sessionJsonl": ".fixtures/runs/propose-graph-commit/2026-06-04-existing-code-ref/session.jsonl",
+    "transcriptMarkdown": ".fixtures/runs/propose-graph-commit/2026-06-04-existing-code-ref/transcript.md",
+    "reportJson": ".fixtures/runs/propose-graph-commit/2026-06-04-existing-code-ref/report.json"
+  }
+}