Wiki Aggregate

Wiki Aggregate — N raw sources → 1 structured pack

A protocol for agentic aggregation of long-horizon research material. Inverts the standard "concat all → ask LLM to summarize" pipeline: instead, an aggregator agent navigates the N source files with three tools, building a notes scratchpad with full path:line provenance, and finally writes a structured pack (brief / findings / sources / aggregation log).

Core principle: don't read everything upfront. Don't merge final answers. Treat the N sources as a queryable environment.

Why this exists

Three traditional ways to aggregate N parallel research outputs all fail on long-horizon, open-ended tasks:

  ❌ concat all sources into one prompt
       → 200K+ token explosion, attention collapse on long context

  ❌ summarize each, then merge summaries
       → ~90% of intermediate evidence (the "I noticed X but..." asides) is lost

  ❌ LLM-as-judge picks the single best source
       → discards the other N-1 sources' independent findings

These failure modes show up clearly on open-ended research tasks where there's no ground-truth verifier. The alternative: treat the N sources as an environment, send a lite agent in to inspect / search / synthesize on demand. Cost ≈ a single rollout, recall is materially higher, and cross-source contradictions get surfaced explicitly.

This skill is the protocol. No Python, no MCP — pure markdown protocol that any harness with Read + Grep can execute.

When to Use

• You have 3-30 raw notes / transcripts / research dumps on one topic, scattered across files
• The current pack on this topic feels thin or you suspect raw evidence wasn't fully lifted
• You want claims with path:line provenance, not vibes
• You want cross-source contradictions surfaced, not buried

When NOT to Use

• N < 3 sources — protocol overhead exceeds value, just write a summary by hand
• N > 30 — budget will be exhausted; narrow the scope or batch by sub-topic
• Sources are not yet collected — this is a synthesis tool, not a research tool. Run your collectors first.
• You only need a one-shot Q&A — use a reactive wiki-ask-style skill instead

The Aggregation Loop

                    Trajectories-as-environment
            ╔════════════════════════════════════════╗
            ║                                        ║
            ║   src_1   src_2   src_3  ...   src_N   ║
            ║   [..]    [..]    [..]         [..]    ║
            ║   [..]    [..]    [..]         [..]    ║
            ║                                        ║
            ╚═══════════════════╤════════════════════╝
                                │
                                │  not concatenated.
                                │  not summarized.
                                │  navigated.
                                │
                                ▼
            ┌────────────────────────────────────────┐
            │       AGGREGATOR (lite agent)          │
            │  ┌──────────────────────────────────┐  │
            │  │  inspect_file / inspect_section  │  │
            │  │  search_sources                  │  │
            │  │  cross_pack_check                │  │
            │  └──────────────────────────────────┘  │
            │                                        │
            │  scratch state:                        │
            │   notes   = []   # {claim, evidence,   │
            │                  #  source, line_ref}  │
            │   budget  = 25   # tool calls          │
            │   subtopics = derived from skim pass   │
            │                                        │
            │  loop until: subtopic coverage met,    │
            │              OR budget = 0,            │
            │              OR 2 zero-info calls      │
            └───────────────────┬────────────────────┘
                                │
                                ▼
                ┌─────────────────────────────┐
                │   pack/                     │
                │     brief.md                │
                │     findings.md   ← claims  │
                │     sources.tsv   ← S-IDs   │
                │     _aggregation_log.md     │
                └─────────────────────────────┘

Process

Phase 1: Scope

1. Resolve the source glob. Count N.
2. Hard stop if N < 3 — abort and tell the user to either collect more or summarize manually.
3. Warn if N > 30 — suggest narrowing by sub-topic or batching across runs.
4. If updating an existing pack, locate it and read its current brief.md + findings.md so you know what already exists.

Phase 2: Cheap-Pass (Skim)

For each source, do one cheap read:

• File ≤ 200 LOC: read the whole file.
• File > 200 LOC: read first 80 lines (frontmatter + intro + section headers).

Build an in-memory source map:

S1  | path/to/source_1.md  | what it covers (1-2 lines) | rough_topics
S2  | path/to/source_2.md  | ...                        | ...

This pass costs ~N reads, each bounded. Do not skip — the source map is what makes Phase 3 efficient.

Phase 3: Aggregator Loop (budget = 25 by default)

Tool inventory (use whatever your harness provides — Read, Grep are sufficient):

Loop discipline:

state.notes = []
state.budget = 25  (or user-specified)

while state.budget > 0:
    pick highest-value next action:
      drill         — a subtopic has a hot lead in one source
      cross_search  — a claim from S1 should be cross-checked against others
      dedup_check   — a claim looks novel; verify no existing pack covers it
      resolve       — two sources disagree; inspect both passages
      explore       — a subtopic has zero notes after Phase 2; broaden search
      DONE          — coverage threshold met

    record note → {claim, evidence_quote, source_id, line_ref, confidence}
    state.budget -= 1

Stopping criteria — declare DONE when ANY holds:

• ≥3 high-confidence notes per subtopic
• Budget exhausted
• Two consecutive tool calls returned no new info

Hard rule: every note MUST have a source_id + line_ref (path + line range). No provenance, no claim. This is what makes the pack auditable.

Phase 4: Write the Pack

Output location: <pack-name>/. If updating, merge with existing files (preserve original sources for existing claims, add new claims, flag superseded ones).

Files:

1. brief.md — 200-400 word executive overview. Subtopic skeleton. Reading order suggestion.

2. findings.md — claims, one block per finding, grouped under subtopic headers:

   ## Claim: <one-line claim>
   Status: supported | contradicted | uncertain
   Confidence: high | medium | low
   Sources: S1, S3, S7
   Evidence:
   - "exact quote or paraphrase" — S1 (path/to/source.md:L120-128)
   - "..." — S3 (path/to/other.md:L45-50)
   Notes: <optional — e.g., "S3 contradicts S7 on date">

3. sources.tsv — S-ID mapping:

   id   path                          type             captured_at   url_or_origin
   S1   path/to/source_1.md           podcast-notes    2026-04-12    https://...
   S2   path/to/source_2.md           daily-intel      2026-04-13    ...

4. _aggregation_log.md — always written. Audit trail:

   # Aggregation Log
   Date: YYYY-MM-DD
   Topic: <topic>
   Sources: N=<N>
   Tool calls: X / budget Y
   Cross-pack overlaps: <list or "none">
   Subtopics covered: <list>
   Skipped sources (no relevant content): <list>
   Stopping criterion triggered: <which one>