BB knowledge-compiler
Knowledge compiler: teaches agents to incrementally compile source files into persistent wikis for cross-session knowledge accumulation. Runtime dependencies: Python 3.8+ (stdlib + pydantic). Optional enhancements: WebSearch (active search), external MCP validator (logic validation). Five modes, automatically routed based on user intent: recall → User wants to answer questions based on existing knowledge. Trigger words: recall, knowledge mode, open wiki, answer with knowledge, according to wiki, based on accumulation, check wiki, does wiki have, researched before, previously organized. ingest → User provided new material, wants to compile into wiki. Trigger words: ingest, compile, organize this, digest this, study this, archive, add this in, accumulate, research this, help me organize, add to knowledge base. query → User asked a specific question, wants to find answer from wiki (one-time). Trigger words: query, answer according to wiki, what does wiki say, check it. lint → User wants to check wiki health. Trigger words: lint, check wiki, wiki health, clean up, any conflicts. deep-dive → User wants Agent to automatically find knowledge gaps and fill them. Trigger words: deep-dive, deep dive, fill gaps, research gaps, level up, auto-complete wiki, knowledge completion, comprehensive fill. Note: deep-dive is not an independent mode, it's a combined pipeline of lint(Coverage) + ingest(search-fill).
As a process B 65/100 · Nearly there — weak spots: when it triggers, consistency
How to improve
- Shorten the description to 1024 characters.
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 0
✓ No critical or high findings
Files scanned: 18. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- error
description-longdescription is 1424 chars, limit 1024
Process rating: all ten parameters 65/100
- 20When it triggers. No condition that starts the skill
- 40Consistency. Frontmatter name (knowledge-compiler) differs from the folder (auto-wiki)
- 60Tools and files. Uses tools (bash, web, python) that frontmatter does not declare
- 60Result and completion. Output format stated, no completion criterion
- 70Inputs and preconditions. Inputs and preconditions are listed
- 70Failures and branches. 5 branches
- 70Execution cost. Instruction body is 4335 tokens
- 100Steps. 43 steps
- 100Running it twice. Mutating operations check current state
- 100Progress reporting. Reports progress
- low 11 top-level sections: this looks like several domains in one skill
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +5Description has no quoted example phrases that should trigger the skill
- +4Description does not say when NOT to use the skill (false activations)
- +3Description length 1423: 120–800 characters recommended
- +1No license
- +2Single-language instructions
- +4Structure: 21 headings
- +3Step-by-step instructions: 43 items
- +3Output format is stated explicitly
- +4Has examples (11 code blocks)
- +4Reference files are cited in the instructions (9 of 9)
Quality base 70; lint remarks subtract, signals add up to 100. Result: 63.