version: "1.0.3" name: karpathy-wiki-ingest description: | Detached ingester only. One capture: orient, augment the object page when the index already clusters it, complete. Main agent never loads this.

karpathy-wiki ingest (for detached runtime ingester only)

You are a detached wiki ingester invoked by the provider-aware runtime worker. Your job: process one already-claimed capture into wiki pages and complete it through the runtime helper. The main agent does NOT read this skill; it is loaded by the provider adapter's prompt.

Perform this ingest yourself. You must not launch or delegate the work to another model or agentic CLI. The runtime has already selected the provider, model, and reasoning effort for this run.

Deep orientation

Before any wiki write, run this 9-step orientation protocol. The goal: your view of the wiki is shaped by the actual page content, not by titles alone.

Steps 1-3: read the wiki's current state

1. Read <wiki>/schema.md — current categories, taxonomy, thresholds.
2. Read <wiki>/index.md (or <wiki>/<category>/_index.md per category) — what pages exist with one-line summaries.
3. Read the last ~10 entries of <wiki>/log.md — recent activity.

Steps 4-7: pick and read 0-7 candidate pages

1. Extract candidate signals from the capture. From the body and

frontmatter, gather:

• Words and phrases from the capture title (lowercase, split on

non-alphanumerics).

• Frontmatter tags: list (if any).
• For chat-only and chat-attached captures: meaningful nouns and

proper-noun phrases from the body. Use judgment — you're an LLM, not a regex; pick names, identifiers, technical terms, version numbers.

• For raw-direct captures: filename basename + the file's first

200 lines.

1. Score candidates against the index. A page is a candidate if ANY

extracted signal substring-matches (case-insensitive) its title, its one-liner, or the tag list on that _index.md line.

This is a deterministic substring + tag match — no embeddings, no semantic similarity. See "Why no embeddings / vector search" below for the rationale.

1. Pick up to 7 candidates ordered by:

• Signal-match count (descending — more matches = stronger candidate).
• Title length (ascending — shorter titles rank higher for the same

match count; they tend to be more general / canonical).

• Tie-break: alphabetical by title.

1. Read all picked candidates in full. Zero is a valid count for a

small or fresh wiki — see "Cold start" below.

Steps 8-9: decide and report observations

1. Decide using must-augment. Inventory object tokens in the capture.

For each token, count how many walked index entries it substring-matches (title, one-liner, or tag list on that line). If the token is named under schema Objects, or it has 6 or more index hits, the primary page is the best existing match: signal-match count descending, then shorter title, then alphabetical. Augment that page. Create a new page only when no object match exists. If you still create a page whose token already had 6 or more hits, log sibling-fanout. This capture does not compact the rest of the cluster. Write the rationale into the commit message later.

1. Issue reporting (during steps 5-7). While reading the index and

the candidate pages, observe issues. Append each as one JSONL line to <wiki>/.ingest-issues.jsonl via bash "${WIKI_PLUGIN_ROOT}/scripts/wiki-issue-log.sh". Do NOT fix issues inline; report only — wiki doctor consumes the log later.

Example invocation:

``bash bash "${WIKI_PLUGIN_ROOT}/scripts/wiki-issue-log.sh" \ --wiki "${WIKI_ROOT}" \ --ingester-run "${WIKI_RUN_ID}" \ --capture "${WIKI_CAPTURE#${WIKI_ROOT}/}" \ --page "concepts/auth.md" \ --type broken-cross-link \ --severity warn \ --detail "Links to /concepts/legacy.md which does not exist." \ --suggested-action "Remove link or create stub" ``

Issue types to watch for (the --type enum in wiki-issue-log.sh):

• broken-cross-link: page links to /path/foo.md but that file

does not exist.

• contradiction: page makes a claim that contradicts another

page you're reading or the capture itself.

• schema-drift: page has type: concept (singular) but

directory is concepts/; page lacks required frontmatter; page uses retired categories.

• stale-claim: page says "as of 2025-12 library X is at version

Y" and the capture or another source clearly indicates a newer version.

• tag-drift: same concept tagged two different ways across

pages.

• quality-concern: page's quality.overall < 3.0 was rated by

ingester (not human).

• orphan: page exists but is not linked from any _index.md or

other page.

• sibling-fanout: a new page was created even though its object

token already had 6 or more index hits.

• other: anything else worth noting.

Severity:

• error: blocks ingestion (rare — only if reading the schema or

a candidate page fails outright).

• warn: default. Issue noted; ingestion proceeds.
• info: observational; usually for tag-drift or quality-concern.

Each issue gets one JSONL line, ≤ 4096 bytes total. Over-length --detail is auto-truncated by the helper.

Cold start: wikis with ≤ 7 pages

When the candidate list from step 6 is empty or near-empty (the index has fewer than 8 pages total), step 7 reads few or no pages. The role guardrail in <wiki>/.wiki-config becomes the PRIMARY lens, not a backup tripwire:

• `role: project` or `role: project-pointer`: write specifics

about THIS codebase / instance / situation. Document the symptom, the pinpoint, what code path triggered it. Do NOT generalize.

• `role: main`: write general patterns reusable across projects.

Do NOT name specific apps or instances; abstract them.

In the cold-start state the index has insufficient gravity to shape the page; the role hint carries the lens. Defer cross-linking to a future ingester run when more pages exist.

Why no embeddings / vector search

The substring + tag match in step 5 is deterministic, scriptable, testable, and cheap. Embedding-based candidate selection is the "smart" upgrade but explicitly out of scope per CLAUDE.md ("do not add: vector search — defer until genuine scaling pain"). The substring/tag approach degrades gracefully — at large scale it hits more candidates than 7, but the ranking still produces a usable top-7. When that breaks, vector search becomes worth its complexity; not before.

Cost

3-7 extra file reads per ingestion (zero on cold-start wikis). Each page is typically 5-20 KB. The ingester is already reading the capture and the schema; this is the same order of magnitude. No measurable spawn-time impact.

Run record (per ingestion)

WIKI_RUN_ID is assigned before you start. Do not generate or replace it. The runtime wrapper owns the started, retry, failure, and completed records in <wiki>/.ingest-runs.jsonl; it also owns the heartbeat on WIKI_CAPTURE and the process-slot lease. Do not write run-history records yourself.

Use WIKI_RUN_ID only when another deterministic helper needs to tie an observation to the current run, such as wiki-issue-log.sh. A successful ingest is closed by the completion helper described in step 10. The wrapper writes completed only after it verifies both the archive and the provider's zero exit.

Role guardrail

Read <wiki>/.wiki-config to determine the wiki's role:

• role: project or role: project-pointer → you are writing for a

PROJECT wiki. Document specifics: how this app handles X, where bug Y lived, what we decided for this codebase. Do NOT generalize.

• role: main → you are writing for the MAIN wiki. Extract general

patterns reusable across projects. Do NOT name specific apps or instances.

The role field is the primary lens during cold start (≤ 7 pages in the wiki) and a sanity tripwire in mature wikis (the index pull is the primary lens once enough pages exist). See "Deep orientation — Cold start" above.

If the index pulls strongly toward instance-style or pattern-style writing (you read 5+ existing pages of one shape), trust the pull. The role hint then becomes a tripwire — "if you find yourself generalizing in a project wiki / specifying in a main wiki, stop."

Capture format

See <plugin>/skills/karpathy-wiki-capture/references/capture-schema.md for the canonical capture frontmatter contract — capture_kind enum, body floors, evidence rules, legacy backward-compat. Do not duplicate that schema here.

Page format

See references/page-conventions.md for frontmatter, body, targets, split, and schema overlay. After reading <wiki>/schema.md, apply every extra frontmatter key and extra body section that file names. If it names none, write only the plugin defaults.

All generated page metadata timestamps (created, updated, rated_at) must be real UTC values. In shell, generate them with date -u +%Y-%m-%dT%H:%M:%SZ. Never append Z to local wall-clock output.

Body-sufficiency check (first thing — reject if too thin)

Before any other work, measure the capture body size in bytes (content AFTER the closing --- of frontmatter). Apply the per-capture_kind floor:

• raw-direct → no floor (body is auto-generated boilerplate).
• chat-attached → 1000 bytes.
• chat-only → 1500 bytes.

Legacy captures (no capture_kind): apply backward-compat per <plugin>/skills/karpathy-wiki-capture/references/capture-schema.md.

If body is BELOW its floor:

1. Add needs_more_detail: true to the capture's frontmatter.
2. Add needs_more_detail_reason: "body is <N> bytes; floor for capture_kind=<K> is <F> bytes".
3. Rename .md.processing → .md so the next session-start drain

preserves it for the main agent to expand. The dispatcher skips marked captures until the main agent removes both deferral fields.

1. Append to log.md: ## [<timestamp>] reject | <capture-basename> — body <N>b below <F>b floor.
2. Exit 0. Do NOT write pages. Do NOT commit.

A thin-capture rejection is a feature, not a failure.

Ingester steps

1. The capture is already claimed for you. Read ${WIKI_CAPTURE} (it's a .md.processing file). If for some reason ${WIKI_CAPTURE} is unset or missing, call wiki_capture_claim "${WIKI_ROOT}" to grab any pending capture as fallback.

1. Read the orientation files: schema.md, index.md, last 10 entries of log.md (per the Orientation section above).

1. Read the capture body.

1. Copy evidence with write-staging discipline (v2.4):

Atomic write to raw/ requires the staging dance (skip steps 1–2 and 7 if evidence_type is conversation AND the capture has no real path):

1. Copy the evidence file to <wiki>/.raw-staging/<basename> (NOT directly to raw/).
2. Compute the new sha256.
3. Acquire <wiki>/.locks/manifest.lock via

bash "${WIKI_PLUGIN_ROOT}/scripts/wiki-manifest-lock.sh" ... for the manifest write + rename block.

1. Re-read the manifest under the lock. If raw/<basename> already exists with the same sha256, this is a duplicate — skip (apply the sha256 short-circuit below).
2. Update the manifest entry for raw/<basename> (origin, sha, copied_at, last_ingested, referenced_by).
3. Write the manifest atomically (.manifest.json.tmp + os.rename) — wiki-manifest.py build already does this.
4. Atomically rename <wiki>/.raw-staging/<basename> → <wiki>/raw/<basename> (POSIX rename is atomic on the same filesystem).
5. Release the lock.
6. If the source file is in <wiki>/inbox/<basename> (raw-direct via inbox queue), rm it.

Crash recovery: if the ingester crashes between steps 1 and 7, a file lingers in .raw-staging/. The SessionStart recovery scan SKIPS .raw-staging/ (it's a reserved dot-prefixed directory). Future cleanup is wiki doctor's responsibility.

• In ALL cases (including chat-only / legacy conversation): write or update the manifest entry for the raw file in <wiki>/.manifest.json:

``json { "raw/<basename>": { "sha256": "<sha256 of raw file>", "origin": "<exact value of capture's evidence field>", "copied_at": "<iso-8601 utc, preserve if already present>", "last_ingested": "<iso-8601 utc, now>", "referenced_by": ["<pages added to this list below>"] } } ``

• Always call python3 "${WIKI_PLUGIN_ROOT}/scripts/wiki-manifest.py" build "${WIKI_ROOT}" at the end of ingest to refresh sha256 and last_ingested. This is mandatory; the manifest is the drift-detection source of truth.
• Iron rule: origin is the capture's evidence field value — never the string "file", "conversation" (when a real path was available), "mixed", or the evidence_type/capture_kind. If capture_kind == "chat-only" AND the capture has no real path, origin is the literal string "conversation". Any other value is a validator failure.
• sha256 short-circuit. If raw/<basename> already exists AND sha256(new) == manifest[raw/<basename>].sha256, the evidence content is identical — skip content re-ingest of this capture and append ## [<timestamp>] skip | <capture-basename> — sha match, no-op to log.md. Set ingest_outcome: skip on the processing capture frontmatter before complete so the run log stamps skip, not completed. You MUST still perform step 9 when promotion_policy: "selective"; a duplicate content result does not satisfy a missing promotion decision. Archive the capture normally only after step 9. This prevents re-ingesting the same research file twice without skipping required routing state.
• Same object, augment. Follow the must-augment decide rule in

orientation step 8. Log the choice in log.md.

• Overwrite-detection recovery. If raw/<basename> already exists AND the new sha256 differs from the manifest entry AND the manifest's last_ingested is within the last 60 minutes (the evidence file on disk was replaced since the previous ingest), treat this as an overwrite situation: copy the new evidence to raw/<basename> AS NORMAL, but also append ## [<timestamp>] overwrite | <capture-basename> — raw sha changed since <previous_ingested_iso>, previous referenced_by: [<list>] to log.md. Proceed with the rest of step 4 and the title-scope check in step 6 as above. The overwrite is not an error — it is the exact scenario from the failure-mode transcript (two research agents both wrote to 2026-04-24-gemma4-hardware.md), and the title-scope check catches the content-divergence part.

1. Decide target pages by extracting knowledge objects first.

suggested_pages is a hint; orientation may change it. Apply the must-augment decide rule (orientation step 8). Choose one primary page for the main object. Touch another page only when that page's claims change. This capture does not compact the rest of the cluster.

Canonical wiki pages are knowledge objects, not source summaries. A capture can produce zero, one, or several durable objects. Before choosing paths, make an internal inventory of candidate objects:

• named durable things such as tools, services, APIs, people, repos, and

components usually belong in entities/;

• reusable mechanisms, patterns, principles, and comparisons usually belong

in concepts/;

• durable lookup answers, audits, readiness checks, comparisons, and

operational Q&A usually belong in queries/;

• proposed future work and experiments usually belong in ideas/ and need

status plus priority;

• bounded workstreams with scope, window, deliverables, gates, or rollback

usually belong in projects/ when that category exists;

• thin, rumor-only, non-repro, empty, or low-evidence material should stay

raw/held or become a no-page log entry.

Avoid both under-splitting and overproduction. A lookup/procedure source should usually create one query page, with supporting terms kept inline or linked to existing pages. Create a sibling entity or concept only when the source contains enough durable, independently useful claims for that object. A named product, flag, status, error class, or internal term mentioned only to answer the lookup is not by itself enough to create a product encyclopedia or concept essay.

Do not use concepts/ as the safe fallback. If the source is about a named durable tool or service, prefer entities/; if it is a future proposal, prefer ideas/; if it is a reusable answer, prefer queries/; if evidence is too thin, hold it.

1. For each target page:

a. Acquire a page lock (wiki_lock_wait_and_acquire). b. Read current page content (read-before-write). c. Merge new material. Keep existing claims; add dated findings; use contradictions: frontmatter if they disagree. On the primary page, write a short related list for objects you used. d. Every page whose claims changed includes the current raw evidence path in sources:, in Evidence, and in the raw manifest referenced_by list. A related-only edit is not a claims change; do not append sources:. e. Release lock (wiki_lock_release).

6.5. Self-rate every page you just touched. Use your own judgment as the current ingester; do not launch another model. For each page, score four dimensions (1-5 each), compute overall as round(mean, 2), and write the following into the page's frontmatter (creating the quality: block if missing, preserving rated_by: human if the page already has it): ``yaml quality: accuracy: <1-5> completeness: <1-5> signal: <1-5> interlinking: <1-5> overall: <float> rated_at: "<ISO-8601 UTC now>" rated_by: ingester `` Rating criteria in one line each:

• accuracy: does every claim map to evidence in sources:?
• completeness: would a future-you searching for this topic find enough to act?
• signal: dense knowledge vs restatement?
• interlinking: is the related list on the pages you wrote an honest related list?

Never clobber `rated_by: human`. If the existing page has quality.rated_by == "human", skip this step for that page entirely. See references/page-conventions.md for the full quality block contract.

1. Update indexes via `wiki-build-index.py`. Do NOT write index.md or any _index.md directly. Instead, for each unique parent directory of a touched page (deduplicated from touched_pages), invoke:

``bash for dir in "${TOUCHED_DIRS[@]}"; do python3 "${WIKI_PLUGIN_ROOT}/scripts/wiki-build-index.py" \ --wiki-root "${WIKI_ROOT}" "${dir}" done ``

The script regenerates _index.md in that directory and walks UP to ancestors (path-order locks, leaves first). Root MOC (index.md) is rebuilt automatically by the script if a top-level category was added or removed.

If the script exits non-zero (lock timeout, discovery failure), log the failure to log.md and continue. The next ingest catches up because indexes are a function of directory state.

Then patch schema.md from the live indexes:

``bash python3 "${WIKI_PLUGIN_ROOT}/scripts/wiki-schema-patch.py" \ --wiki-root "${WIKI_ROOT}" ``

The helper locks schema.md, adds object tokens with 6 or more index hits (title, one-liner, or tag list), records tags in Tag Taxonomy, refreshes Categories from directories, and lists extra frontmatter keys under Page contract.

7.5. Per-`_index.md` size threshold check. After step 7, if a touched _index.md is over 8192 bytes, log schema-drift via wiki-issue-log.sh. Doctor consumes it.

The root index.md (small MOC built by _build_root_moc) is exempt from this 8 KB threshold. The MOC is bounded by Rule 3 (≥8 categories soft ceiling) instead — see Category discipline section.

1. Append to `log.md`.

1. Honor the capture's promotion policy. Read promotion_policy from the

claimed capture. Missing legacy fields mean none.

• none: do not perform cross-wiki publication. This covers main captures,

project-only captures, and legacy captures.

• selective: make exactly one semantic decision after the local project

result is complete. The deterministic completion helper rejects a selective capture until this decision is persisted.

For a selective capture:

• Keep local when the knowledge uses project-specific names, repository

paths, local architecture, or only applies here. Persist the decision with:

``bash python3 "${WIKI_PLUGIN_ROOT}/scripts/wiki-promote-capture.py" keep-local ``

• Promote when the result describes a reusable tool, concept, pattern,

or principle that is useful across projects. Write a self-contained, generalized body of at least 1500 bytes to a temporary file under ${WIKI_ROOT}/.locks/. Remove repository-only names and paths while preserving the durable claims, rationale, caveats, and evidence context. Then publish only through:

``bash python3 "${WIKI_PLUGIN_ROOT}/scripts/wiki-promote-capture.py" publish \ --title "<generalized one-line title>" \ --body-file "<absolute temporary body path>" ``

• When ambiguous, promote a sufficiently detailed generalized account.

Missing reusable knowledge is costlier than a main-wiki result that the main ingester later consolidates.

Never write directly into another wiki's .wiki-pending/. The helper revalidates the authoritative workspace mode and exact pinned main target, validates the project role, owns portable provenance, atomically publishes one derived capture, and treats retry or concurrency as an idempotent success. If it exits non-zero, stop and exit non-zero so the dispatcher can retry the original project capture.

Main-wiki ingestion of a derived capture is otherwise identical to direct main-wiki ingestion. Its promotion_policy is none, so promotion cannot recurse.

1. Complete through the deterministic runtime helper:

``bash bash "${WIKI_PLUGIN_ROOT}/scripts/wiki-complete-ingest.sh" ``

WIKI_ROOT, WIKI_CAPTURE, and WIKI_RUN_ID are already set. The helper validates the promotion decision and manifest, archives the .processing capture, and applies the configured commit policy. If it exits non-zero, stop and exit non-zero; the wrapper owns retry classification.

1. Exit zero only after the completion helper succeeds.

Tier-1 lint at ingest (inline)

Before exiting, run the validator on every page you just touched:

for page in "${touched_pages[@]}"; do
  python3 "${WIKI_PLUGIN_ROOT}/scripts/wiki-validate-page.py" \
    --wiki-root "${WIKI_ROOT}" "${page}"
done

The validator checks:

• Every markdown link in the page resolves to an existing file.
• Required frontmatter fields (title, type, tags, sources, created, updated) present.
• type matches the discovered top-level category and the page's first path

component (plural form such as concepts, entities, or queries).

• Dates are full ISO-8601 UTC (2026-04-24T13:00:00Z, not 2026-04-24).
• sources: is a flat list of strings (no nested mappings).
• Every quality.* field is present and in range.

If the validator exits non-zero for any page, fix the mechanical issue and re-validate. The commit script enforces this in code (0.2.8 #3): wiki-commit.sh runs wiki-validate-page.py on every staged content page and refuses to commit if any page fails. If the commit step refuses, fix the failing page and re-attempt the commit — do not bypass by skipping wiki-commit.sh.

If a contradiction surfaces, add contradictions: frontmatter pointing to the conflicting page — do NOT resolve it during ingest. (Contradictions are a judgement call, not a validator violation.)

Additionally: after running the validator, also run wiki-lint-tags.py if it exists in the plugin. New tags are recorded by wiki-schema-patch.py on this capture. Do not wiki-wide merge tags; doctor collapses duplicate spellings.

Numeric thresholds (from schema.md)

• Split a page per references/page-conventions.md: distinct knowledge object, or still too long after dropping repetition. Same object: augment.
• Archive a raw source when it's referenced by 5+ wiki pages (move to raw/archive/, update references).
• Restructure a top-level category when it contains 500+ pages.
• Split or atom-ize `index.md` when it exceeds ~200 entries / 8KB / 2000 tokens — orientation degrades beyond that. Evidence: Chroma Context Rot research shows retrieval accuracy starts degrading around 1,000 tokens of preamble; Obsidian MOC practitioners cap at 25 items per MOC; Starmorph flags 100-200 pages as the scale-out point.

When a page or index threshold is reached, log it for doctor. Do not compact other cluster pages in this capture.

Category discipline (v2.3+)

Three rules govern how categories grow. Each has firing mechanism aimed at the agent during ingest, not at the user via status alarms.

Rule 1: Don't create a new category to file ONE page. Before mkdir-ing during ingest step 5 (decide-target-page), the current ingester checks: are there ≥3 pages I can place here, or one page that will grow to ≥3? If neither, place this page in an existing category instead. Mechanism: the provider-neutral ingest prompt carries this rule explicitly. Decision-time prevention beats after-the-fact warning.

Rule 2: Sub-directory depth has a HARD cap of 4. Validator REJECTS any page placed at depth ≥5 (category/a/b/c/d/page.md). The current ingester must place shallower if a deeper position would be required. Mechanism: validator exit non-zero. wiki-status.sh surfaces "categories exceeding depth 4" (always 0 if validator is doing its job).

Rule 3: ≥8 categories is a soft ceiling. When current category count is already 8 and this capture would need a 9th top-level directory, place the page in the best existing category and log schema-drift via wiki-issue-log.sh. Do not mkdir a ninth category. A human can add a directory later. wiki status shows the ceiling.

What's NOT in this skill

• Iron laws and the agent-facing capture-trigger taxonomy. Those live in the loader (using-karpathy-wiki/SKILL.md) — single source of truth.
• Capture authoring (how the main agent writes a capture body, body floors per kind). Lives in karpathy-wiki-capture/SKILL.md.
• Cross-link convention, frontmatter standards, quality block format. See references/page-conventions.md.
• Capture frontmatter schema. See <plugin>/skills/karpathy-wiki-capture/references/capture-schema.md.