pstack-claude-mirror 0.0.0-stage → 0.9.133

package/plugins/pstack/skills/why/references/synthesizer-prompt.md

@@ -0,0 +1,130 @@
+# Synthesizer Prompt Template
+
+Fill in the placeholders.
+
+---
+
+You are answering a "why" question about a piece of code by synthesizing findings from investigators who searched different historical sources (source control, issue / ticket tracker, long-form documents, real-time team chat, infrastructure observability, error / exception tracking, product analytics warehouse, and code comments). Produce a confidence-weighted, evidence-cited narrative that says honestly what the evidence supports, and just as honestly what it doesn't.
+
+## The Question
+
+> {QUESTION}
+
+## The Code Anchor
+
+**Target files:** {FILES_WITH_LINE_RANGES}
+
+**Key symbols:** {SYMBOLS}
+
+## Investigator Findings
+
+{ALL_INVESTIGATOR_FINDINGS}
+
+## Sources That Weren't Searched
+
+{SKIPPED_SOURCES_WITH_REASONS}
+
+## Epistemics Framework
+
+You MUST follow `references/epistemics.md`. Read it in full before writing. The key rules:
+
+1. Every claim must sit in one tier: **Direct**, **Supported**, **Inferred**, **Speculative**, **Unknown**. The tier sets its section and its phrasing.
+2. Every Direct/Supported claim must have a citation (PR #, ticket ID, doc URL, chat permalink, commit hash, or file:line).
+3. Inferred and Speculative claims must use hedged language ("appears to", "likely", "suggests", "one possibility is").
+4. Never cite code as evidence for its own intent.
+5. Document gaps. Don't fill them with plausible-sounding guesses.
+6. Treat a hypothesis embedded in the user's question as a candidate, not a conclusion. Check it independently.
+
+## Instructions
+
+1. **Read all investigator findings** and weigh the evidence yourself. Investigators gathered raw evidence, not conclusions.
+2. Apply [Claims about human decisions](epistemics.md#claims-about-human-decisions) to claims about human rulings. Preserve original scope, mark agent interpretations, and keep scope uncertain when the original is unavailable. Repeated copies of one summary are one evidence chain.
+3. **Merge overlaps.** When several investigators cite the same PR, ticket, or doc, give it one authoritative reference.
+4. **Surface contradictions.** When two items disagree, show both. Don't pick one.
+5. **Calibrate each claim.** Find its evidence and tier. Direct: cite it and state it plainly. Inferred: hedge it and show the inference. Speculative: mark it. No evidence: put it in the gaps.
+6. **Verify citations by spot-checking.** If you're uncertain a cited item exists or says what the investigator claims, check it by reading the codebase or calling MCP tools. Don't propagate errors. Do not write files, commit, or modify external state.
+7. **Don't overreach.** The user will act on this. Leave an open question open rather than fill it with a confident guess.
+
+## Output Format
+
+Write for the user, in this exact structure:
+
+---
+
+### The Question
+
+Restate the user's question in one or two sentences.
+
+### The Code in Question
+
+File paths, line ranges, key symbols. Two or three lines to orient a cold reader.
+
+### What We Found
+
+Claims with textual evidence, each quoted or paraphrased and cited precisely:
+
+- **[Direct]** {Claim}. Source: [PR #123](https://example.com/pr/123) / ticket ID / file:line. {Brief quote or paraphrase.}
+- **[Supported]** {Claim}. Evidence: {list of items and what each contributes}.
+
+`[Direct]` is single-source, explicit evidence. `[Supported]` is multiple indirect items converging on a conclusion.
+
+### What We Can Reasonably Infer
+
+Claims stated nowhere but well-supported by indirect evidence. Each bullet must make the inference chain visible ("Given A and B, it's likely that C") and must use hedged language ("appears to", "likely", "suggests", "is consistent with"):
+
+- **[Inferred]** {Hedged claim}. Reasoning: {the specific evidence and the inference step}.
+
+Skip this section if there's nothing to infer.
+
+### Competing Hypotheses
+
+If the evidence fits multiple stories, present each. Don't force a winner the record doesn't support. For each hypothesis:
+
+- **Hypothesis:** {one-sentence statement}
+- **Evidence for:** {specific items}
+- **Evidence against or missing:** {what would need to be true but isn't, or what counter-signals exist}
+
+Skip this section if there's a single clear answer.
+
+### What We Don't Know
+
+Specific gaps: "We searched the issue tracker for [query1], [query2], [query3] and found no issue discussing the rate-limit threshold", not "We don't know why." Include:
+
+- Questions that went unanswered
+- Searches that returned nothing
+- Sources that were unavailable, and why (such as a missing real-time team chat MCP)
+- People who would likely know but whom you can't ask
+
+### Sources Consulted
+
+What was actually searched, so the user can judge coverage and redirect:
+
+- **Source control history**: {file paths}, {number of commits reviewed}, PRs #{numbers}, and code comments searched. Or "Not searched. This should not happen because git and `gh` are always expected."
+- **Issue / ticket tracker**: {ticket IDs and keyword searches}. Or "Not searched. No matching MCP available in this environment."
+- **Long-form documents**: {page titles and search queries}. Or "Not searched. No matching MCP available in this environment."
+- **Real-time team chat**: {channels searched, date ranges, queries}. Or "Not searched. No matching MCP available in this environment."
+- **Infrastructure observability**: {dashboards, monitors, metrics, logs, traces, or incidents searched}. Or "Not searched. No matching MCP available in this environment."
+- **Error / exception tracking**: {issues, events, or releases searched}. Or "Not searched. No matching MCP available in this environment."
+- **Product analytics warehouse**: {fully-qualified tables queried, the time windows, and the numeric summaries (counts, percentiles, first/last-seen timestamps) that bore on the question}. Or "Not searched. No matching MCP available in this environment."
+
+### Confidence Summary
+
+One or two sentences on overall confidence, e.g.:
+
+> "The core rationale (A) is well-supported by direct PR and ticket evidence. The specific threshold value (100) is inferred from the surrounding context but not explicitly documented. The question of whether this was driven by a customer request could not be answered. No relevant issue tracker or long-form doc content surfaced, and real-time team chat search was unavailable."
+
+---
+
+## Quality Check Before Returning
+
+Revise if any item fails:
+
+1. Does every claim in "What We Found" have a citation? If not, add one or move it to "Inferred" or "Hypotheses."
+2. Is the phrasing tier-appropriate? (Direct claims can use "because". Inferred claims cannot.)
+3. Did you surface every contradiction you noticed, or quietly pick one?
+4. Does "What We Don't Know" name specific gaps? If it's empty or missing, be suspicious. Historical investigations almost always have gaps.
+5. Did you check any hypothesis embedded in the question against the evidence rather than rubber-stamp it?
+6. Did you cite code as evidence for its own intent? Remove it. Code is mechanics, not motivation.
+7. Is the tone calibrated? A confident-sounding answer on weak evidence is the exact failure this skill exists to prevent.
+
+The value of this output is its honesty, not its authority. A reader who takes it to the original author, an engineering lead, or a product manager should know what's known, what's inferred, and what's missing, so they can ask the right follow-up questions. Optimize for being useful, not for looking decisive.

package/tools/forks.json

@@ -0,0 +1,398 @@
+{
+  "pstack": {
+    "plugins/pstack/agents/comment-sicko.md": {
+      "kind": "policy",
+      "why": "Sends a comment that states a dependency, platform, vendor, or protocol rule to that system's documentation or source code, makes each kill of one cite the page or file and list it in the report, and marks the comment HOLD when that hunt cannot settle the rule, where the keep list alone would delete it on doubt.",
+      "since": "0.9.106",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/architect/SKILL.md": {
+      "kind": "policy",
+      "why": "Requires comparing accepted deviations with the saved sketch and rationale, and updating them, before each implementation unit closes.",
+      "since": "0.9.34",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/architect/references/rationale-template.md": {
+      "kind": "policy",
+      "why": "Adds an Implementation reconciliation section that records accepted deviations and their acceptance sources.",
+      "since": "0.9.34",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/blast-radius/SKILL.md": {
+      "kind": "policy",
+      "why": "Extends the trigger to run before design when a brief asserts something about existing code.",
+      "since": "0.9.21",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/bro/SKILL.md": {
+      "kind": "policy",
+      "why": "Adds a /bro and say-it-plainly trigger to the description.",
+      "since": "0.9.10",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/create-verification-skill/SKILL.md": {
+      "kind": "port-feature",
+      "why": "Names the generated project skill verify so it replaces Claude Code's bundled user-only /verify and the playbooks can invoke it.",
+      "since": "0.9.29",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/how/references/explainer-prompt.md": {
+      "kind": "policy",
+      "why": "Tells a lone explainer on the simple path that no explorers ran and that it traces the code itself with Read, Grep, and Glob before explaining.",
+      "since": "0.9.130",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/no-comments/SKILL.md": {
+      "kind": "policy",
+      "why": "Treats a HOLD that names its dependency, platform, vendor, or protocol rule and where to check it as settled for the run, restores as a HOLD any kill of such a comment without a refuting page or file, and limits a second rejected report's failure to that report's findings so the accepted work still lands. Sets comment-sicko's model from the no-comments reviewer line in pstack-models.md.",
+      "since": "0.9.106",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-help/SKILL.md": {
+      "kind": "port-feature",
+      "why": "Routes installation, model-sheet loading, and persistent routing preferences to the current runtime's setup sources, gives installed playbooks precedence over upstream guides, and keeps the port's public skill loading, hidden principles, verification-skill naming, local-worker advice, driver and built-in equivalents, excluded make-bot-ui row, and orchestrate wording as prose changes in this fork, and routes \"stack them, don't ship\" to Autopilot-stack, because neither autopilot merges.",
+      "since": "0.9.71",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/SKILL.md": {
+      "kind": "policy",
+      "why": "Adds the Platform Adaptation section, whose runtime sentence the generator stamps from RUNTIMES, with its task-tool and todo.md fallback (per task slug when sessions share a checkout, and beside the desktop app's short update_plan list), the run-or-verify driver line, the deploy-platform and defect-severity triggers, the port's principle summaries, the project playbooks paragraph, and the closed-review-gate rule that the Codex and Pi mappings point to, and it points the Agent-call defaults at the Models section with full tool access and the namespaced pstack:poteto-agent name, requires stopping an abandoned agent before replacing it, renders every reply item as prose, layers technical-writing on top of unslop, and describes Autopilot-full as stopping at merge-ready for the operator's click and Autopilot-stack as Autopilot-full for PRs that build on each other.",
+      "since": "0.9.2",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/playbooks/autopilot-full.md": {
+      "kind": "policy",
+      "why": "Owners stop at merge-ready and the operator clicks every merge, with the patch-id rule governing re-verification if trunk moves first, instead of each owner merging after a clean verdict, and the root writes the program objective into the store's notes.md on the go and re-reads it each tick, arms the tick as a dynamic-mode /loop, states the program's size before the first owner starts and waits for the operator's go above three owners, scopes review lanes to what the diff breaks, and stops a PR at two fix-forwards.",
+      "since": "0.9.39",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/playbooks/autopilot-stack.md": {
+      "kind": "policy",
+      "why": "Records each verdict's head SHA, base SHA, base branch, and patch-id, cancels pending merges and leases each push against the captured remote head before any restack, re-verifies a verdict whose applicability to the new base is uncertain, and lists verdicts in place of the merge-ready list, because Autopilot-full already stops at merge-ready.",
+      "since": "0.9.34",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/playbooks/babysit.md": {
+      "kind": "policy",
+      "why": "Requires confirming that the first status read reports the requested PR or stack, and naming the merge frontier, before acting on it, and reports the watcher's behind-base gate to the branch owner instead of resolving it, as it does a conflict.",
+      "since": "0.9.39",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/playbooks/bug-fix.md": {
+      "kind": "policy",
+      "why": "The reply quotes the decisive failing and passing output trimmed to the assertion and counts, not the whole repro verbatim.",
+      "since": "0.9.10",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/playbooks/feature.md": {
+      "kind": "policy",
+      "why": "Gives every file-writing delegate its own worktree, based on the branch's HEAD when it builds on unmerged commits, and forbids writing in a worktree a delegate still holds.",
+      "since": "0.9.39",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/playbooks/multi-phase-plan.md": {
+      "kind": "port-feature",
+      "why": "Plans read skills from the installed plugin, keep the store under ~/.claude/orchestrate, write the program objective into the store's notes.md, run the tick as a dynamic-mode /loop that re-reads that objective, run live lanes in worktrees through the driver skill the Non-negotiables resolve instead of on cloud VMs with control-ui or control-cli, name the configured swarm workers model instead of a placeholder, leave rebase and merge authority to the execution playbook, and pick Autopilot-stack only when a PR builds on another, because neither autopilot merges.",
+      "since": "0.9.19",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/playbooks/opening-a-pr.md": {
+      "kind": "policy",
+      "why": "Requires stopping every live agent that holds a worktree before committing, merging, or deploying from it.",
+      "since": "0.9.10",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/playbooks/pause-safely.md": {
+      "kind": "policy",
+      "why": "Writes the resume note to the canonical checkpoint directory and preserves artifacts the user asked to keep across the resume.",
+      "since": "0.9.34",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/playbooks/session-pickup.md": {
+      "kind": "policy",
+      "why": "Reads the project checkpoint first and applies the human-rulings epistemics when reconstructing state, and verifies inherited claims before it routes and ends.",
+      "since": "0.9.34",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/playbooks/shipping.md": {
+      "kind": "policy",
+      "why": "Lands one PR at a time under the user's landing authority, cancelling pending merges before any chain change, merging only under a service-enforced head condition, arming auto-merge or a queue only behind repository rules that enforce the verdict, and confirming each landing in the destination's history before advancing, with verifiers that drive the real surface through the driver skill in their own worktrees, a matching patch-id that keeps a review only after its applicability to the new base is checked, and the watcher owning blocker classification, with a branch behind its base routed to the topology owner.",
+      "since": "0.9.34",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/playbooks/worktree-cleanup.md": {
+      "kind": "policy",
+      "why": "Holds a worktree with untracked files or a git lock instead of calling untracked files safe to drop, names the untracked count and the lock reason the audit prints, says that a worktree with both tracked edits and untracked files prints both counts, names the status call the audit counts from and a diff command that covers submodules, says that an empty diff never means nothing is at stake because a diff leaves untracked files out and a plain one can leave a submodule out, says that a wip line can stand for work inside a submodule, limits the prune step's kept-commits claim to commits on the worktree's branch and says a submodule's commits go with the worktree, holds ignored files outside the audit's disposable build and cache directories and commits only the worktree's HEAD reflog reaches, naming the ignored and reflog counts and the commands that list them, and forbids rm -rf around a locked tree.",
+      "since": "0.9.74",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/bootstrap.ts": {
+      "kind": "policy",
+      "why": "Exits with a clear message when run without bun and installs production dependencies only.",
+      "since": "0.9.11",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/check-plan.mjs": {
+      "kind": "port-feature",
+      "why": "Checks for the port's plan skeleton markers (the objective written into the program store's notes.md, installed-plugin reads, an hourly cadence, the configured swarm workers model), requires a box under each program task and under Close the program, excludes fenced examples from required section and review-gate text, opens and closes a code fence at any indentation so a fence inside a list item counts, and strips a leading byte-order mark before parsing.",
+      "since": "0.9.19",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/check-plan.test.ts": {
+      "kind": "port-feature",
+      "why": "Gives upstream's valid-plan fixture the boxes under each program task and under Close the program, the notes.md, installed-plugin, and hourly lines, and the configured swarm workers lane line that the port's checker requires.",
+      "since": "0.9.83",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/package.json": {
+      "kind": "policy",
+      "why": "Typechecks through the root tsconfig.json so the port-only modules are covered.",
+      "since": "0.9.23",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/watch-pr/cli.test.ts": {
+      "kind": "policy",
+      "why": "Supplies the shared deadline and landing revision the port's watcher requires, and checks that a PR with no checks configured gets a first-sighting status cell and, after one WAITING poll, a READY verdict that names the absence, and that mergeability GitHub has not computed is named in the status table, the WAITING line, and the TIMEOUT line, that a retry after a mid-read change is not called a failed query, that a command that could not run gets an install action, that an --interval longer than one timer can sleep is refused, and the two help lines the port rewrote.",
+      "since": "0.9.34",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/watch-pr/cli.ts": {
+      "kind": "policy",
+      "why": "Threads one WatchDeadline through discovery, commands, and polling, reports a deadline verdict instead of throwing, rejects an --interval longer than one timer can sleep, and says in its help that --interval does not shorten the 60 second no-checks confirmation and that exit 0 from --status-only does not mean ready.",
+      "since": "0.9.24",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/watch-pr/fakes.test-helper.ts": {
+      "kind": "policy",
+      "why": "Extends the fake reader with the base revision, the default branch, and a factsOnReread option for every pullRequest read after the first, parses its facts through the real parsePullRequest, and tags each rollup page, repeating the last one across polls, so a test can replay a head commit with no rollup.",
+      "since": "0.9.10",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/watch-pr/github.test.ts": {
+      "kind": "policy",
+      "why": "Covers the port's GitHub reader: completed conclusions that gh buckets as pending, a checkout PR paired with a different explicit repository, a full open-PR page refused as a possibly cut stack, a stack that never takes a PR headed at the default branch as parent, child, or seed under the default branch the repository reports, while a fork branch of the same name still stacks, a check-rollup cursor that repeats or returns to an earlier page, a review-thread cursor that repeats, a review-thread page that promises more pages without a cursor, a binary that cannot be spawned, the base revision field, and a no-checks reading that needs both check reads to agree and fails closed otherwise. Expects the `missing-field` query failure.",
+      "since": "0.9.24",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/watch-pr/github.ts": {
+      "kind": "policy",
+      "why": "Runs every gh and git call under the shared deadline, reports a binary that cannot be spawned as a non-retryable query failure, reads the base revision, pages through every review thread, refuses a review-thread or check-rollup cursor that does not advance and a review-thread page that promises more pages without a cursor, fails completed conclusions that gh buckets as pending, refuses a checkout PR paired with a different explicit repository, reads the repository's default branch, leaves every PR whose head is that branch in the PR's own repository out of stack ordering as parent, child, and seed, orders a stack only through branches in the PR's own repository, while rejecting duplicate head branches, cycles, and a full open-PR page, and reads a repository with no checks configured as its own result when gh reports none and the head commit has no rollup. Throws the `missing-field` query failure.",
+      "since": "0.9.24",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/watch-pr/policy.test.ts": {
+      "kind": "policy",
+      "why": "Covers the review-required and branch-protection merge gates and their wait for pending checks, the behind-base gate that stops without waiting for pending or unreported checks and outranks a required review but not requested changes, the wait while GitHub has not computed mergeability (never ready and never a query error, re-polled at the interval, timing out at the caller's deadline, one status row on a one-shot pass, and blockers that do not depend on it still stopping at once), a PR with no checks configured (waiting in single, stack, and queued mode until the reading has persisted for 60 seconds however short the interval, each PR timed from its own first sighting and unconfirmed when the caller supplies no confirmer, timing out there under a shorter deadline, ready once it has, still stopped by conflicts, threads, and gates, honouring checks that register on the next poll, and failing closed while the head may not have reported), the second facts read in the snapshot plan and the retry when the PR's state, merge time, head branch, gate, review, draft, or conflict state changes between the two reads, except mergeability that the first read had not computed, and the shared deadline in the poll loop.",
+      "since": "0.9.24",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/watch-pr/policy.ts": {
+      "kind": "policy",
+      "why": "Stops at the review-required and branch-protection merge gates once checks finish, and at the behind-base gate at once and ahead of a required review, since updating the branch restarts its checks, waits at the poll interval while GitHub has not computed mergeability, never as a query error, and builds a ready proof only from a MERGEABLE reading, re-reads the PR facts after the checks and retries a snapshot whose head, base, mergeability, review, or draft state changed in between, leaving mergeability out of that comparison when the first read had not computed it, reads review threads and checks concurrently, reports a PR with no checks configured as ready only when no commit on it has reported checks and the same head has shown none for 60 seconds of wall time whatever the poll interval, waiting until then as it waits on pending checks, and honours the shared deadline in the poll loop. Names the changed-fact and wait-key locals `fact` and `waitKey`, not `key`, because the plugin directory's scanner reads a `key` identifier beside a file read as a credential read (MCP_FORWARDS_CREDENTIAL_ENV).",
+      "since": "0.9.24",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/watch-pr/render.ts": {
+      "kind": "policy",
+      "why": "Renders the review-required, behind-base, and merge-blocked states as status cells and as merge-gate blockers with their next action, and shows a PR with no checks configured as a no-checks cell and a checks=none line on its READY verdict, with a first sighting as its own cell, WAITING line, and TIMEOUT line, and mergeability GitHub has not computed as its own Merge cell, WAITING line, and TIMEOUT line, words a retry after the PR changed mid-read as that and not as a failed query, and tells a caller whose command could not run to install it.",
+      "since": "0.9.43",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/watch-pr/types.compile.ts": {
+      "kind": "policy",
+      "why": "Adds the landing revision to the compile-time type fixture and pins that a first sighting of no checks is not readiness proof.",
+      "since": "0.9.34",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/poteto-mode/scripts/watch-pr/types.ts": {
+      "kind": "policy",
+      "why": "Adds the review-required, behind-base, and merge-blocked gate reasons, the landing revision on open PR facts and in the ready proof, the head repository of an open PR, the default-branch reader method, deadline, spawn-failed, snapshot-changed, and invalid-stack query failures, the mergeability-unknown wait reason, and the no-checks reading (a none-reported fast path, a no-rollup page, the ci-none state, the ci-unreported first sighting, and the checks-unreported wait reason). Names the query failure for a missing response field `missing-field`, not `missing-key`, for the same scanner.",
+      "since": "0.9.24",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/principle-build-the-lever/SKILL.md": {
+      "kind": "policy",
+      "why": "Requires executable semantic controls for each claimed comparison output and refusal or inconclusive results when required observations are missing.",
+      "since": "0.9.91",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/principle-prove-it-works/SKILL.md": {
+      "kind": "policy",
+      "why": "Adds checks that the process behind a result was sound, that a failing check fails for the predicted reason, and that a success report rests on the real state and exit code rather than truncated output.",
+      "since": "0.1.0",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/principle-test-behavior-not-implementation/SKILL.md": {
+      "kind": "policy",
+      "why": "Counts a constant pin as behavior when the value is a promise to users, such as a documented default.",
+      "since": "0.9.82",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/recall/SKILL.md": {
+      "kind": "policy",
+      "why": "Sweeps the locations the project's standing rules name before reporting anything as never recorded and applies the human-rulings epistemics, on top of Claude Code's transcript path.",
+      "since": "0.9.2",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/reflect/SKILL.md": {
+      "kind": "port-feature",
+      "why": "Finds the session transcript with the port's find-transcript.mjs under the Claude Code projects directory that claude-project-dir.mjs prints, keeps the parent's conclusions out of the reviewer prompts (a fallback digest records events without a verdict, and nothing goes beside a transcript path), and drops Cursor's readonly agent mode and inline model column.",
+      "since": "0.1.0",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/setup-pstack/SKILL.md": {
+      "kind": "port-feature",
+      "why": "Writes the override sheet at each runtime's own path, loaded by @-include on Claude Code, pasted into AGENTS.md on Codex, and read directly by the Muse plugin hook since none has Cursor's always-applied rules, replaces Cursor's budget step with a default effort and per-role effort levels and explains Codex's active-turn metadata requirement for session effort, adds the SessionStart hook toggle and names the Pi extension and OpenCode plugin as readers of it, keeps every runtime's sheet path table and per-runtime step changes in references/runtimes.md, routing GitHub Copilot to copilot.md, Kimi and Grok to runtime-specific setup, and OMP to native model configuration and its global routing toggle, and numbers its verification and formal verification offers as steps 9 and 10.",
+      "since": "0.9.2",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/show-me-your-work/SKILL.md": {
+      "kind": "policy",
+      "why": "Documents that log.sh also neutralizes a cell opening with a double quote.",
+      "since": "0.9.41",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/show-me-your-work/scripts/log.sh": {
+      "kind": "policy",
+      "why": "Prefixes a cell opening with a double quote so a spreadsheet or TSV reader cannot parse it as a quoted field, appends each row through one perl syswrite so parallel writers cannot interleave rows longer than the shell's stdio buffer, reports a short or failed write, and falls back to upstream's printf when perl is missing or does not work.",
+      "since": "0.9.41",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/backends/cfg.ts": {
+      "kind": "policy",
+      "why": "Reads the section keyword from `keyword`, not `key`, because the plugin directory's scanner reads a `key` identifier beside a file read as a credential read (MCP_FORWARDS_CREDENTIAL_ENV).",
+      "since": "0.9.128",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/frontends/config.ts": {
+      "kind": "policy",
+      "why": "Names the section keyword `keyword`, not `key`, because the plugin directory's scanner reads a `key` identifier beside a file read as a credential read (MCP_FORWARDS_CREDENTIAL_ENV).",
+      "since": "0.9.128",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/frontends/messages.ts": {
+      "kind": "policy",
+      "why": "Drops the \\r that TLC on Windows leaves at the end of each line, so an end-of-message line closes its message there too.",
+      "since": "0.9.116",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/frontends/spec.ts": {
+      "kind": "policy",
+      "why": "Reads the section keyword from `keyword`, not `key`, because the plugin directory's scanner reads a `key` identifier beside a file read as a credential read (MCP_FORWARDS_CREDENTIAL_ENV).",
+      "since": "0.9.128",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/frontends/view.ts": {
+      "kind": "policy",
+      "why": "Names the index signatures `field`, not `key`, because the plugin directory's scanner reads a `key` identifier beside a file read as a credential read (MCP_FORWARDS_CREDENTIAL_ENV).",
+      "since": "0.9.128",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/ir.ts": {
+      "kind": "policy",
+      "why": "Names the section keyword `keyword`, the value-key function `valueKey`, and the map-helper and radix-sort locals `k` and `rank`, not `key`, because the plugin directory's scanner reads a `key` identifier beside a file read as a credential read (MCP_FORWARDS_CREDENTIAL_ENV).",
+      "since": "0.9.128",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/passes/label.ts": {
+      "kind": "policy",
+      "why": "Imports `valueKey` and names the per-state key array `keyAt`, not `key`, because the plugin directory's scanner reads a `key` identifier beside a file read as a credential read (MCP_FORWARDS_CREDENTIAL_ENV).",
+      "since": "0.9.128",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/passes/regions.ts": {
+      "kind": "policy",
+      "why": "Names the per-state key array `keyAt`, not `key`, because the plugin directory's scanner reads a `key` identifier beside a file read as a credential read (MCP_FORWARDS_CREDENTIAL_ENV).",
+      "since": "0.9.128",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/run.ts": {
+      "kind": "policy",
+      "why": "Kills the worker's process tree with taskkill on Windows, which has no /bin/sh and no process groups. Gives mise and java a named environment through childEnv, not the whole environment, because the plugin directory's scanner reads a whole environment given to a child beside a download URL as a credential forward (MCP_FORWARDS_CREDENTIAL_ENV).",
+      "since": "0.9.116",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/statechart.ts": {
+      "kind": "policy",
+      "why": "Spawns janitor.ts on Windows in place of the /bin/sh janitor, points the worker's TEMP and TMP at its temp folder as TMPDIR already is, and hides the windows of the detached worker and janitor.",
+      "since": "0.9.116",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/test/chart.test.ts": {
+      "kind": "policy",
+      "why": "Imports `valueKey`, names the fixture's record variable `lock`, and keeps the word out of its test titles, because the plugin directory's scanner reads a `key` identifier beside a file read as a credential read (MCP_FORWARDS_CREDENTIAL_ENV).",
+      "since": "0.9.128",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/test/cli.test.ts": {
+      "kind": "policy",
+      "why": "Resolves the temp folder before the symlinked-launcher test computes its relative link, because macOS keeps that folder under the /var symlink and the link would otherwise miss the launcher, so the suite passes on macOS. The signal-while-writing test's poll returns false when its walk meets the worker's rename of draft to done, instead of failing with ENOENT. Spawns the command with the named child environment from run.ts, not the whole environment, for the same scanner.",
+      "since": "0.9.105",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/test/config-fuzz.test.ts": {
+      "kind": "policy",
+      "why": "Names the section keyword `keyword` and the index signature `field`, not `key`, because the plugin directory's scanner reads a `key` identifier beside a file read as a credential read (MCP_FORWARDS_CREDENTIAL_ENV).",
+      "since": "0.9.128",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/test/corpus.test.ts": {
+      "kind": "policy",
+      "why": "Spawns the command with the named child environment from run.ts plus TLA2TOOLS and JAVA, not the whole environment, because the plugin directory's scanner reads a whole environment given to a child as a credential forward (MCP_FORWARDS_CREDENTIAL_ENV).",
+      "since": "0.9.128",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/statechart/scripts/test/messages.test.ts": {
+      "kind": "policy",
+      "why": "Adds a test that TLC's output with Windows line ends reads as it does with Unix ones.",
+      "since": "0.9.116",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/tdd/SKILL.md": {
+      "kind": "policy",
+      "why": "Requires quoting the failure content, not the assertion line, as evidence that the new test fails for the intended reason.",
+      "since": "0.9.21",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/why/references/epistemics.md": {
+      "kind": "policy",
+      "why": "Adds the Claims about human decisions section that five other files cite for keeping a ruling's original wording and scope.",
+      "since": "0.9.34",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/why/references/investigator-prompt.md": {
+      "kind": "policy",
+      "why": "Tells investigators to return a human ruling's attributable original wording per the epistemics section.",
+      "since": "0.9.34",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/why/references/synthesizer-prompt.md": {
+      "kind": "policy",
+      "why": "Adds a step that applies the human-rulings epistemics, and gives the Direct-claim example a URL the link validator resolves.",
+      "since": "0.9.17",
+      "upstream": "not-proposed"
+    }
+  },
+  "cursor-team-kit": {
+    "plugins/pstack/skills/deslop/SKILL.md": {
+      "kind": "policy",
+      "why": "Scopes the diff to the PR's base branch, else the branch it was cut from, else the default branch, with uncommitted changes, and reruns the verifying checks after editing, reverting any edit whose check fails.",
+      "since": "0.9.130",
+      "upstream": "not-proposed"
+    }
+  },
+  "thermos": {
+    "plugins/pstack/skills/thermo-nuclear-code-quality-review/SKILL.md": {
+      "kind": "policy",
+      "why": "Adds a Report Format section requiring one summary report plus one detail file per subsystem or reviewer.",
+      "since": "0.9.11",
+      "upstream": "not-proposed"
+    },
+    "plugins/pstack/skills/thermos/SKILL.md": {
+      "kind": "port-feature",
+      "why": "Sets both reviewers' model from the thermos reviewers line in pstack-models.md, so the override sheet reaches them.",
+      "since": "0.9.115",
+      "upstream": "not-proposed"
+    }
+  }
+}