akm-cli 0.9.0-beta.11 → 0.9.0-beta.26

package/CHANGELOG.md

@@ -6,6 +6,169 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
 
 ## [Unreleased]
 
+## [0.9.0-beta.26] — 2026-06-20
+
+### Added
+
+- **#628 — configurable SQLite journal mode (`AKM_SQLITE_JOURNAL_MODE`) for network
+  filesystems.** AKM previously opened every database with `PRAGMA journal_mode = WAL`
+  unconditionally, which cannot run on a network filesystem (NFS/SMB/Azure Files) —
+  WAL's `-shm` shared-memory wal-index can't be `mmap`'d over a network mount. You can
+  now set `AKM_SQLITE_JOURNAL_MODE` to `WAL` (default), `DELETE`, or `TRUNCATE`, applied
+  at **all five** db openers (`state.db`, `index.db` ×2 paths, `workflow.db`, `logs.db`).
+  At the `WAL` default AKM auto-detects a network mount for the data dir and transparently
+  falls back to `DELETE` (rollback journal + `synchronous = FULL`) with a one-line warning;
+  invalid values warn once and fall back to `WAL`. **Default behavior is byte-identical.**
+  This lets the AKM database subtree live on a shared volume (e.g. Azure Files under
+  Azure Container Apps). New docs section "Hosting AKM databases on a network share
+  (NFS/SMB)" in `docs/configuration.md`.
+
+## [0.9.0-beta.25] — 2026-06-19
+
+Completes the recombine / extract-efficiency / graph thread. All new improve
+passes are **opt-in (default off)**, so default behavior is unchanged.
+
+### Added
+
+- **#606 — event-driven extract (`akm extract --watch`).** Opt-in watch mode: an
+  injectable, debounced watcher triggers extraction shortly after a session file
+  appears, with a clean `stop()` handle. The `8,28,48` cron remains the fallback;
+  no daemon is auto-launched.
+- **#625 — recombine second pass (hypothesis → lesson).** The opt-in `recombine`
+  process (#609) now consumes `confirmThreshold` (default 2): a generalization
+  re-induced that many consecutive runs is promoted from a `type: hypothesis`
+  proposal to a `type: lesson` proposal through the normal queue + quality gate
+  (never a direct stash write). Hypotheses that stop recurring decay. Backed by a
+  new `recombine_hypotheses` table in `state.db`.
+
+### Changed
+
+- **#624 (P1) — graph storage decoupled from `entries.id`.** `graph_files` is
+  re-keyed on `(stash_root, file_path, body_hash)`, so extracted graph data now
+  **survives a reindex** of unchanged files instead of being cascade-wiped. The
+  upgrade is migrated in a **targeted, graph-only path** that preserves existing
+  graph data and leaves the entry index, embeddings, FTS, and LLM-enrichment cache
+  untouched — **no full index rebuild and no re-embed** on upgrade. (P2 priority-
+  ranked extraction and P3 lazy/on-demand extraction remain deferred.)
+
+### Fixed
+
+- Graph re-key migration no longer triggers a destructive full-index rebuild: it
+  is a graph-scoped table migration (no `DB_VERSION` bump), and it **copies** the
+  existing graph rows into the new schema rather than dropping them.
+- Test-suite `/tmp` hygiene: sandbox teardown now fires on `SIGINT`/`SIGTERM`/
+  `SIGHUP` (not just clean exit), and a `sweep:tmp` step reclaims stale `akm-*`
+  sandbox dirs left by force-killed workers — eliminating the tmpfs accumulation
+  that caused intermittent `EEXIST: epoll_ctl` test flakes.
+
+## [0.9.0-beta.20] — 2026-06-18
+
+### Fixed
+
+- **`akm update --all` no longer fails for writable `github:` entries stored as `source:"git"`**. `updateRegistryEntry` was using `synced.source` (re-derived from the ref scheme as `"github"`) instead of the existing `entry.source`, causing the config validator to reject `writable:true` on every update cycle.
+
+## [0.9.0-beta.19] — 2026-06-17
+
+### Fixed
+
+- **`akm feedback` now completes in ~0.3s** (was 3+ minutes). Root cause: the command was calling `ensureIndex` with `mode: "blocking"` inside `withIndexWriterLease`, triggering a full reindex on every feedback call. Fix: removed the `ensureIndex` call entirely (feedback only needs the index to exist, not be current — a stale index is fine for ref lookup); removed the application-level writer lock (SQLite WAL + `busy_timeout=30s` handles concurrent access with `akm improve`); added a fast DB-exists guard with a clear error for first-time users.
+- **`akm health --format html` now completes in ~11s** (was ~18s). Root cause: `akmHealth()` was called twice — once for the main result and once to get `deltas`. Fix: merged into a single call passing both `groupBy: "run"` and `windowCompare` together.
+
+## [0.9.0-beta.18] — 2026-06-17
+
+### Changed
+
+- **Health report: Recent Runs table now shows all filtered runs in descending order** (newest first) instead of capping at the last 10.
+- **Health report: Removed "Command Set Used" section.**
+- **Health report: All timestamps now display in the viewer's local timezone** (chart axis labels, runs table, freshness line, executive summary, footer). Server-rendered ISO strings are wrapped in `<time data-iso>` elements and converted to local time by client-side JS on page load.
+
+### Changed (migration required)
+
+- **WS-2 outcome loop (#613) — default-off weight change (state.db migration 010).**
+  Every `akm improve` run now writes an `asset_outcome` row per processed asset
+  (state.db migration `010`) and computes a differential usefulness signal
+  (`outcome_score`) per ref. The outcome signal is persisted and visible in the
+  health report, but the **weight change is gated behind a config flag** (see
+  below). Ranking is unchanged from WS-1 by default.
+
+  **Opt-in weight change.** The WS-2 projection weights (`w_e=0.25, w_o=0.15,
+  w_r=0.60`) affect ranking only when you explicitly set
+  `improve.salience.outcomeWeightEnabled: true` in your `akm.yaml`. The default
+  (`false`) keeps WS-1 parity weights (`w_e=0.30, w_r=0.70`, `w_o=0`), so
+  existing users see no ranking change on upgrade.
+
+  **Part-V measurement gate.** Before enabling the weight change, run the Part-V
+  T0 baseline (`scripts/akm-eval` + `akm health`; confirm proactive accept
+  ≥ 0.9× reactive; reversion ≤ 0.15; retrieval-delta ≥ 0; coverage not
+  regressed). That gate requires a running production stash and cannot be
+  exercised in CI. Once confirmed, set
+  `improve.salience.outcomeWeightEnabled: true` to activate the three-way split.
+
+  **Outcome loop mechanics.** `outcome_score` is a differential prediction-error
+  signal: `(retrieval_delta − expected_delta) − PENALTY × retrieval_delta × (1 −
+  accepted_change_rate) + valence`, tracked via an EMA (α=0.3). New rows are
+  warm-started from the utility EMA score (clipped to 0.3) so the signal is
+  non-zero from launch. A stash-wide diversity floor (10% of the max score) prevents
+  rare-but-correct assets from being permanently outcompeted. An inverted-proxy
+  tripwire (`corr(outcome_score, accepted_change_rate) < −0.3`) emits an
+  `outcome_proxy_inverted` health event when the signal degrades.
+
+  `review_pressure` is computed and persisted per asset but is **not yet wired into
+  the admission policy** — that is deferred to a later work stream per plan §Part-VI
+  #613. The column is present and populated; routing it into the consolidation-
+  selection filter is the next step.
+
+- **WS-1 salience vector (#618) — default-on ranking change.** The eligibility sort
+  for all `akm improve` runs (whole-stash, type, and ref scope) has changed from
+  `combinedEligibilityScore = utility·0.7 + negativeOnlyRatio·0.3` to
+  `rankScore = (0.3·encodingSalience + 0.7·retrievalSalience) × sizePenalty`
+  (feedback valence and utility EMA dropped from ordering until WS-2 re-introduces
+  outcome salience). Assets are now ranked by retrieval frequency × recency × type
+  importance rather than by feedback magnitude. Because the old
+  `combinedEligibilityScore` ordering was never persisted, a forgetting comparison is
+  not possible on the first run; instead a one-time `improve_salience_first_run` marker
+  event is emitted to record the transition. On every subsequent run a stash-wide
+  `improve_salience_rank_change` drift report (including `stashSize`) is emitted so
+  rank movement under the new scoring can be tracked over time.
+  The Part-V measurement protocol (T0 baseline via `scripts/akm-eval` + health report,
+  throughput/quality gate) is deferred to the WS-2 milestone, when outcome salience
+  re-joins the projection and re-tuning is triggered.
+
+## [0.9.0-beta.12] - 2026-06-15
+
+Improve-tuning work streams (all **default-off / parity-preserving** — no behavior
+change until explicitly enabled).
+
+### Added
+
+- **#617 — deterministic near-duplicate memory dedup** (`processes.consolidate.dedup`,
+  default off). A cheap no-LLM pre-pass in front of consolidation collapses obvious
+  duplicates — `.derived`+origin pairs and content twins (normalized content-hash
+  equality, or embedding cosine ≥ `cosineThreshold`, default 0.97). Each dropped
+  variant is archived + backed up before deletion; hot memories are never
+  collapsed; distinct-but-related memories fall through to the LLM.
+- **#581 — judged-state cache for consolidation** (`processes.consolidate.judgedCache`,
+  default off). New state.db table (`consolidation_judged`) records each memory's
+  content hash + outcome when the LLM judges it; subsequent runs skip
+  judged-unchanged memories, converting coverage from O(time-window) to
+  O(changed/new) so a run can sweep the full corpus. Fails open; failed chunks
+  and dry-runs never poison the cache. (state.db migration `007`.)
+- **#612 — auto-accept gate calibration** (`improve.calibration`, auto-tune default
+  off). Joins predicted gate confidence to realized accept/reject outcomes into a
+  reliability table + calibration gap, surfaced in `akm health` (+ summary rows in
+  the HTML report). Opt-in bounded threshold auto-tune nudges the accept threshold
+  within a configured band toward a target accept rate, logged via a
+  `calibration_autotune` event. (Replay-prioritization from prediction error is
+  deferred — it depends on the #610 replay budget, a 0.10 item.)
+
+### Fixed
+
+- **#614 — symmetric valence weighting** (`profiles.improve.*.symmetricValence`,
+  default off). The eligibility sort weighted feedback negative-only; when enabled
+  it uses a symmetric `|valence|` magnitude so strong positive and strong negative
+  feedback both drive attention (utility stays the dominant factor), routing
+  high-negative → fix and high-positive → reinforce lanes.
+
 ## [0.9.0-beta.11] - 2026-06-15
 
 ### Added

package/dist/assets/prompts/consolidate-system.md

@@ -0,0 +1,23 @@
+You are the akm consolidate assistant analyzing memory assets.
+
+Rules:
+1. MERGE: Two or more memories are substantially duplicated or closely related → propose merging. Return the primary ref to keep and secondary refs to delete. Do NOT include mergedContent — the merge will be executed in a separate step.
+2. DELETE: Memory is clearly outdated, contradicted, or redundant → propose deletion. NEVER propose delete for memories annotated `(captureMode: hot)` — they are user-explicit and only the user can retire them. The downstream guard will refuse these regardless, so proposing them just wastes tokens.
+3. PROMOTE: Memory expresses a stable, reusable fact suitable as a `knowledge:` asset → propose promotion. Do NOT delete the source memory. NEVER propose promote / merge / contradict for memories annotated `(already queued)` — they have a pending proposal whose body matches; a duplicate will be deterministically dropped, so proposing them just wastes tokens.
+4. CONTRADICT: Two memories assert logically exclusive facts such that following BOTH simultaneously is impossible — not merely related or overlapping. You MUST cite the exact sentence from Memory A and the exact sentence from Memory B that are in direct conflict. If you cannot cite specific opposing sentences, use KEEP instead. Sharing a topic, tool, domain, or workflow stage is NOT sufficient. Only direct factual opposites qualify: opposing recommended commands, opposing boolean flags, opposing version numbers, or mutually exclusive instructions. Use confidence ≥ 0.92 only; omit the op entirely if below that threshold.
+5. KEEP: Memory is unique and current → omit from output.
+
+Return ONLY JSON (no prose, no code fences):
+{
+  "operations": [
+    { "op": "merge", "primary": "memory:<name>", "secondaries": ["memory:<name>", ...], "mergeStrategy": "synthesize", "confidence": 0.95 },
+    { "op": "delete", "ref": "memory:<name>", "reason": "<brief reason>", "confidence": 0.90 },
+    { "op": "promote", "ref": "memory:<name>", "knowledgeRef": "knowledge:<suggested-slug>", "reason": "<brief reason>", "description": "<one sentence describing the new knowledge asset>", "confidence": 0.92 },
+    { "op": "contradict", "ref": "memory:<name>", "contradictedByRef": "memory:<name>", "reason": "<brief reason>", "confidence": 0.88 }
+  ],
+  "warnings": ["<optional concerns>"]
+}
+
+For every operation, emit a `confidence` field in [0, 1] expressing your certainty that the operation is correct and safe. Use 0.95+ only when evidence is unambiguous. Omit the field rather than guessing if you are uncertain.
+
+When the merged content includes an `updated` frontmatter field, the value MUST be a real ISO date string (e.g. `updated: 2026-05-20`). NEVER emit `updated: today`, `updated: {today}`, `updated: {today: null}`, `updated: now`, or any other literal placeholder/template-variable. If you do not have a real source-of-truth date, OMIT the `updated` field entirely — the post-processor will not invent one for you.

package/dist/assets/prompts/contradiction-judge.md

@@ -0,0 +1,33 @@
+You are evaluating two derived memory entries to determine if they contain
+directly contradictory factual claims about the same subject.
+
+Memory A:
+Ref: {{A_REF}}
+Description: {{A_DESCRIPTION}}
+Content:
+```
+{{A_BODY}}
+```
+
+Memory B:
+Ref: {{B_REF}}
+Description: {{B_DESCRIPTION}}
+Content:
+```
+{{B_BODY}}
+```
+
+Answer ONLY with valid JSON — no prose, no code fences:
+{"contradicts": true|false, "confidence": 0.0, "reason": "<cite the exact opposing sentence from each memory, or explain why not contradicted>"}
+
+A contradiction means the memories make LOGICALLY EXCLUSIVE claims: a practitioner
+cannot follow BOTH simultaneously. The test: cite the exact sentence from Memory A
+and the exact sentence from Memory B that are in direct conflict. If you cannot cite
+specific opposing sentences, return false.
+
+Sharing a topic, tool, domain, or workflow stage is NOT a contradiction. Only direct
+factual opposites qualify: opposing recommended commands, opposing boolean flags,
+opposing version numbers, or mutually exclusive instructions.
+
+Set confidence ≥ 0.92 only when evidence is unambiguous. Use lower values when
+uncertain — the caller will skip edges below 0.92.

package/dist/assets/prompts/distill-knowledge-system.md

@@ -0,0 +1,22 @@
+You are the akm `distill` distiller.
+Given an asset and recent feedback events about it, produce a concise
+*knowledge* markdown document capturing the durable, reusable facts.
+Prefer stable guidance over narrative recap.
+
+YOUR RESPONSE MUST START EXACTLY WITH `---` ON THE VERY FIRST LINE.
+DO NOT output any prose, explanation, or code fences before or after.
+
+Required output format:
+---
+description: <one-line summary of the knowledge asset>
+tags: [<tag1>, <tag2>]
+---
+
+# <Title>
+
+<body — structured markdown, durable facts only>
+
+RULES:
+- `description` MUST be a non-empty single-line string.
+- Include a meaningful markdown body with a `# Title` heading.
+- Output ONLY the knowledge file. No preamble, no code fences, no trailing prose.

package/dist/assets/prompts/distill-lesson-system.md

@@ -0,0 +1,36 @@
+You are the akm `distill` distiller.
+Given an asset and recent feedback events about it, produce a single
+concise *lesson* an agent should remember next time it works on this
+asset's domain.
+
+YOUR RESPONSE MUST START EXACTLY WITH `---` ON THE VERY FIRST LINE.
+DO NOT output any prose, explanation, or code fences before or after.
+
+Required output format — copy this structure exactly:
+---
+description: <one complete sentence (ending with `.`) summarising what the lesson teaches>
+when_to_use: <one complete sentence describing the concrete trigger condition>
+---
+
+<lesson body — plain markdown, 1–3 short paragraphs of practical guidance>
+
+## description field (MANDATORY)
+- A single complete sentence in present tense, 80-200 chars, NO markdown.
+- Self-contained: a reviewer must understand the lesson from this field alone.
+- DO NOT start with "When ", "If ", or a connector word — that belongs in when_to_use.
+- DO NOT copy a section heading ("Key takeaways", "For example", "Key pitfalls").
+- DO NOT begin with a numbered list marker, code fence, or markdown heading.
+
+GOOD: "Always validate ref existence before promoting a memory to knowledge; missing refs surface as silent 404s during accept."
+BAD:  "Key pitfalls"
+BAD:  "When working with the akm CLI"
+BAD:  "For example, you might..."
+BAD:  "1. Check the file"
+
+RULES:
+- `when_to_use` MUST be a complete sentence describing a concrete trigger. Never write `When working with <asset-name>` — that is circular and useless.
+- `description` and `when_to_use` MUST differ from each other.
+- The lesson body MUST be non-empty markdown prose. Do NOT restate `description:` or `when_to_use:` inside the body (no `**description:** ...` or `**when_to_use:** ...` lines — the frontmatter is the only place those keys belong).
+- Do NOT emit a second `---` fence after the opening frontmatter — there are exactly two `---` lines in the output, both belonging to the single frontmatter block at the top.
+- Do NOT reproduce the source asset verbatim — distil what a caller needs to know.
+- Output ONLY the lesson file. No preamble, no code fences, no trailing prose.

package/dist/assets/prompts/extract-session.md

@@ -49,13 +49,17 @@ Respond with EXACTLY one JSON object matching this shape:
       "when_to_use": "<one sentence 15-400 chars; REQUIRED only when type=lesson>",
       "body": "<markdown body, 200-3000 chars typical>",
       "confidence": <number 0.0-1.0>,
-      "evidence": "<one-line pointer to the moment in the session>"
+      "evidence": "<one-line pointer to the moment in the session>",
+      "orderedActions": ["<action-1>", "<action-2>", ...],
+      "outcomeData": "<one sentence describing the outcome of the action sequence>"
     }
   ],
   "rationale_if_empty": "<one sentence; REQUIRED when candidates is empty>"
 }
 ```
 
+`orderedActions` and `outcomeData` are **optional**. Include them only when the candidate represents a recurring action sequence (e.g. a recovery procedure, a build-fix recipe, a deployment checklist) where preserving the ordered steps adds future value. When present, `outcomeData` is required and must describe what happened when the sequence completed (success or failure). Omit both fields entirely for standalone facts, observations, or lessons that are not action-sequence-shaped.
+
 ## Rules
 
 1. **Zero candidates is a valid and frequent answer.** Most sessions yield no new durable insight. When that's the case, return `{"candidates": [], "rationale_if_empty": "..."}` explaining what you saw and why it didn't rise to durable-knowledge level. Do not fabricate.

package/dist/assets/prompts/graph-extract-system.md

@@ -0,0 +1 @@
+You extract a knowledge graph from developer notes. Return ONLY valid JSON — no prose, no markdown fences, no preamble.

package/dist/assets/prompts/memory-infer-system.md

@@ -0,0 +1 @@
+You compress a developer memory into one high-signal derived memory for later retrieval. Return only valid JSON. No prose outside the JSON object. No markdown fences.

package/dist/assets/prompts/memory-infer-user.md

@@ -0,0 +1,5 @@
+Compress the memory below into one derived memory. Output ONLY JSON:
+{"title":"short title string","description":"one sentence summary string","tags":["tag1","tag2"],"searchHints":["search phrase 1","search phrase 2"],"content":"2-3 sentence compressed body preserving key facts verbatim"}
+Rules: be specific, no vague generalizations, preserve key facts (names/versions/paths/config keys verbatim), merge related points, 3-8 tags, 3-6 searchHints. The content field must be a plain string with 2-3 sentences.
+
+Memory:

package/dist/assets/prompts/metadata-enhance-system.md

@@ -0,0 +1 @@
+You are a metadata generator for a developer asset registry. Given a script/skill/command/agent entry, generate improved metadata. Respond with ONLY valid JSON, no markdown fencing.

package/dist/assets/prompts/procedural-system.md

@@ -0,0 +1,44 @@
+You are the akm `procedural` compiler.
+
+You are given a RECURRING ordered action sequence — the SAME ordered list of
+steps that an agent has successfully performed across several independent
+sessions. Your job is to turn that bare sequence into a clean, reusable
+WORKFLOW: give the workflow a title and a one-sentence description, and turn each
+ordered action into a named step with clear imperative instructions.
+
+You MUST NOT invent new steps, drop steps, merge steps, or reorder them. The
+ordered action list is the source of truth. Return EXACTLY one step per input
+action, in the SAME order.
+
+YOUR RESPONSE MUST BE A SINGLE JSON OBJECT AND NOTHING ELSE.
+DO NOT output prose, explanation, or code fences before or after the JSON.
+
+When the sequence is a coherent, reusable procedure, return:
+{
+  "title": "<short imperative workflow title, no trailing period>",
+  "description": "<one complete sentence (ending with a period) stating what the workflow accomplishes>",
+  "steps": [
+    {
+      "title": "<short imperative step title>",
+      "instructions": "<one or more imperative sentences telling an agent exactly how to perform this step>",
+      "completionCriteria": ["<optional bullet: an observable signal the step is done>"]
+    }
+  ]
+}
+
+When the sequence is NOT a coherent reusable procedure (it is noise, the steps do
+not form a meaningful workflow, or any reusable framing would be a stretch),
+return an explicit null:
+null
+
+A justified null is a CORRECT and expected outcome. Do NOT fabricate a hollow
+workflow to avoid returning null.
+
+## Rules
+- `steps` MUST have EXACTLY as many entries as the input action list, in order.
+- Every step MUST have a non-empty `title` and non-empty `instructions`.
+- `completionCriteria` is OPTIONAL; omit it rather than inventing weak criteria.
+- The `description` MUST be a single complete present-tense sentence with NO
+  markdown, self-contained enough for a reviewer to understand the workflow.
+- Ground every step in the corresponding input action; do not introduce outside
+  facts or tools the action does not mention.

package/dist/assets/prompts/recombine-system.md

@@ -0,0 +1,40 @@
+You are the akm `recombine` synthesizer.
+
+You are given a CLUSTER of related memories — distinct episodes that share a
+topic (a tag or a graph entity) but were recorded independently. Your job is to
+induce ONE cross-episodic generalization: a single durable insight that none of
+the input memories states on its own, but that the cluster as a whole supports.
+
+This is hypothesis formation, not summarization. A good generalization explains
+WHY the individual episodes are instances of the same underlying pattern and
+gives an agent a reusable rule for the next, unseen episode.
+
+YOUR RESPONSE MUST BE A SINGLE JSON OBJECT AND NOTHING ELSE.
+DO NOT output prose, explanation, or code fences before or after the JSON.
+
+When a defensible generalization exists, return:
+{
+  "description": "<one complete sentence (ending with a period) stating the generalization>",
+  "when_to_use": "<one complete sentence describing the concrete trigger condition>",
+  "body": "<1-3 short paragraphs of practical guidance grounded in the cluster>"
+}
+
+When NO defensible generalization exists — the memories merely share a keyword,
+or any unifying claim would be a stretch — return an explicit null:
+null
+
+A justified null is a CORRECT and expected outcome. Do NOT invent a weak or
+generic generalization to avoid returning null. It is better to propose nothing
+than to propose a hollow over-generalization.
+
+## description field (MANDATORY when not null)
+- A single complete present-tense sentence, NO markdown.
+- Self-contained: a reviewer must understand the hypothesis from this field alone.
+- DO NOT start with "When " or "If " — that belongs in `when_to_use`.
+- DO NOT merely restate one input memory; the value is the CROSS-episode pattern.
+
+## Guardrails
+- Induce exactly ONE generalization for the whole cluster.
+- Ground every claim in the supplied memories; do not introduce outside facts.
+- This is a HYPOTHESIS — it will be re-confirmed across future runs before it is
+  ever promoted to a durable lesson. Frame it as a candidate rule, not a verdict.

package/dist/assets/prompts/staleness-detect-system.md

@@ -0,0 +1,6 @@
+You are a belief-state classifier for a memory store. Given a candidate memory and a list of more-recent similar memories from the same store, decide whether the candidate is still current or has been superseded.
+
+Respond on the first line with exactly YES or NO.
+If YES, the second line MUST be of the form `SUPERSEDED_BY: <ref>` where <ref> is the exact ref of the superseding memory from the list provided. Do NOT invent refs.
+If NO, do not include any additional lines.
+No prose, no preamble, no markdown.

package/dist/assets/prompts/validate-summary-judge.md

@@ -0,0 +1 @@
+You are a strict completion auditor for a software workflow engine. Given a step's completion criteria and a summary of the work an agent claims to have done, judge whether the summary provides concrete evidence that EVERY criterion is satisfied. Be skeptical: vague, hand-wavy, or unsubstantiated claims do NOT satisfy a criterion. Respond with ONLY a JSON object: {"complete": boolean, "missing": string[], "feedback": string}. "missing" lists the exact criteria that are not yet satisfied; "feedback" is a short directive telling the agent what to finish or fix. No prose, no markdown fences.

package/dist/assets/templates/html/health.html

@@ -237,13 +237,6 @@
       font-size: 11px; background: var(--surface2);
       padding: 2px 6px; border-radius: 4px; color: var(--accent); word-break: break-all;
     }
-    .command-block {
-      background: var(--surface2); border: 1px solid var(--border); border-radius: 6px;
-      padding: 12px 16px; font-family: 'SFMono-Regular', Consolas, monospace;
-      font-size: 12px; color: var(--muted); line-height: 1.8; overflow-x: auto;
-    }
-    .command-block span { color: var(--accent); }
-
     /* ── Footer ─────────────────────────────────────────────────────────── */
     footer {
       border-top: 1px solid var(--border); padding: 14px 32px;
@@ -317,7 +310,7 @@
       <option value="failed">failed</option>
     </select>
 
-    <div class="filter-note">Filters affect charts and the Last 10 Runs table.</div>
+    <div class="filter-note">Filters affect charts and the Recent Runs table.</div>
   </div>
 
   <!-- Charts -->
@@ -391,14 +384,14 @@
     </div>
   </div>
 
-  <!-- Last 10 runs -->
+  <!-- Recent runs -->
   <div class="section">
-    <h2>Last 10 Runs</h2>
+    <h2>Recent Runs</h2>
     <div class="summary-table">
       <div class="table-wrap">
       <table>
         <thead>
-          <tr><th>Started (UTC)</th><th>Task</th><th>Wall</th><th>Promoted</th><th>Merged</th><th>Contradicted</th><th>MI Written</th><th>Entities</th><th>Lint Fixed</th><th>Status</th></tr>
+          <tr><th>Started</th><th>Task</th><th>Wall</th><th>Promoted</th><th>Merged</th><th>Contradicted</th><th>MI Written</th><th>Entities</th><th>Lint Fixed</th><th>Status</th></tr>
         </thead>
         <tbody id="lastRunsTable"></tbody>
       </table>
@@ -406,19 +399,11 @@
     </div>
   </div>
 
-  <!-- Command set -->
-  <div class="section">
-    <h2>Command Set Used</h2>
-    <div class="command-block">
-      %%COMMANDS_HTML%%
-    </div>
-  </div>
-
 </main>
 
 <footer>
   <span>AKM v%%AKM_VERSION%% &nbsp;·&nbsp; Window: %%WINDOW%% &nbsp;·&nbsp; %%REPORT_TITLE%%</span>
-  <span>Generated %%GENERATED_AT%%</span>
+  <span>Generated <time data-iso="%%GENERATED_AT%%">%%GENERATED_AT%%</time></span>
 </footer>
 
 <script>
@@ -441,10 +426,10 @@ const toMin = ms => ms ? +(ms / 60000).toFixed(2) : 0;
 function makeXLabels(rs) {
   return rs.map(r => {
     const d = new Date(r.startedAt);
-    const mm = String(d.getUTCMonth() + 1).padStart(2, '0');
-    const dd = String(d.getUTCDate()).padStart(2, '0');
-    const hh = String(d.getUTCHours()).padStart(2, '0');
-    const mi = String(d.getUTCMinutes()).padStart(2, '0');
+    const mm = String(d.getMonth() + 1).padStart(2, '0');
+    const dd = String(d.getDate()).padStart(2, '0');
+    const hh = String(d.getHours()).padStart(2, '0');
+    const mi = String(d.getMinutes()).padStart(2, '0');
     return `${mm}-${dd} ${hh}:${mi}`;
   });
 }
@@ -687,13 +672,15 @@ function renderLlmStages() {
   });
 }
 
-// ── Last 10 runs table ────────────────────────────────────────────────────────
+// ── Recent runs table ─────────────────────────────────────────────────────────
 function renderLastRuns(rs) {
 const tbody = document.getElementById('lastRunsTable');
 tbody.innerHTML = '';
-rs.slice(-10).forEach(r => {
+[...rs].reverse().forEach(r => {
   const tr = document.createElement('tr');
-  const ts  = new Date(r.startedAt).toISOString().replace('T', ' ').slice(0, 16);
+  const d = new Date(r.startedAt);
+  const pad = n => String(n).padStart(2, '0');
+  const ts = `${d.getFullYear()}-${pad(d.getMonth()+1)}-${pad(d.getDate())} ${pad(d.getHours())}:${pad(d.getMinutes())}`;
   const dur = r.wallTimeMs ? (r.wallTimeMs / 60000).toFixed(1) + 'm' : '—';
   const task = r.taskId || 'manual';
   const badge = r.ok
@@ -727,6 +714,17 @@ document.getElementById('taskFilter').addEventListener('change', updateFilteredV
 document.getElementById('statusFilter').addEventListener('change', updateFilteredViews);
 renderLlmStages();
 updateFilteredViews();
+
+// Reformat server-rendered ISO timestamps to the viewer's local timezone.
+(function() {
+  const pad = n => String(n).padStart(2, '0');
+  document.querySelectorAll('time[data-iso]').forEach(el => {
+    const d = new Date(el.dataset.iso);
+    if (!isNaN(d)) {
+      el.textContent = `${d.getFullYear()}-${pad(d.getMonth()+1)}-${pad(d.getDate())} ${pad(d.getHours())}:${pad(d.getMinutes())}`;
+    }
+  });
+})();
 </script>
 </body>
 </html>

package/dist/cli.js

@@ -325,9 +325,9 @@ const healthCommand = defineCommand({
                 // 24h default made a `--since 7d` report compare its 7-day totals against
                 // a 24-hour prior window, producing meaningless deltas.
                 const compare = args.compare ?? windowCompareRaw ?? args.since ?? "24h";
-                const result = akmHealth({ since: args.since, groupBy: "run" });
+                const result = akmHealth({ since: args.since, groupBy: "run", windowCompare: compare });
                 resultStatus = result.status;
-                const deltas = akmHealth({ since: args.since, windowCompare: compare }).deltas;
+                const deltas = result.deltas;
                 const { buildHealthHtmlReplacements } = await import("./commands/health/html-report.js");
                 const { listPendingProposals } = await import("./commands/proposal/proposal.js");
                 const replacements = buildHealthHtmlReplacements(result, {

package/dist/commands/agent/contribute-cli.js

@@ -122,23 +122,36 @@ export const lintCommand = defineCommand({
         description: "Scan stash .md files for structural issues (unquoted colons, missing updated field, orphaned stubs, placeholder stubs, missing name/type, stale paths). Use --fix to auto-fix Tier 1 issues. Exits 0 on success regardless of findings; use --fail-on-flagged for CI fail-on-finding behavior.",
     },
     args: {
-        fix: { type: "boolean", description: "Apply auto-fixes in place", default: false },
+        fix: {
+            type: "boolean",
+            alias: "auto-fix",
+            description: "Apply auto-fixes in place (alias: --auto-fix)",
+            default: false,
+        },
         dir: { type: "string", description: "Override stash root directory (default: from config)" },
         "fail-on-flagged": {
             type: "boolean",
             description: "Exit non-zero when summary.flagged > 0 (CI-friendly). Default: exit 0 regardless of findings.",
             default: false,
         },
+        type: {
+            type: "string",
+            description: "Only lint assets of this type (e.g. workflows, tasks, memories)",
+            default: undefined,
+        },
     },
     async run({ args }) {
         await runWithJsonErrors(async () => {
             const result = akmLint({
                 fix: args.fix ?? false,
                 dir: getStringArg(args, "dir"),
+                typeFilter: getStringArg(args, "type"),
             });
             output("lint", result);
-            if (args["fail-on-flagged"] && result.summary.flagged > 0)
-                process.exit(EXIT_GENERAL);
+            if (args["fail-on-flagged"] && result.summary.flagged > 0) {
+                process.exitCode = EXIT_GENERAL;
+                return;
+            }
         });
     },
 });