CONCEPT Cited by 1 source
Compass, not encyclopedia¶
Definition¶
"Compass, not encyclopedia" is the design principle Meta names for per-module AI-agent context files: each file is 25-35 lines / ~1,000 tokens — short enough that it can be loaded into an agent's context without crowding out tool outputs — with four mandated sections oriented around navigation + failure-avoidance, not exhaustive description.
"No fluff, every line earns its place." (Source: sources/2026-04-06-meta-how-meta-used-ai-to-map-tribal-knowledge-in-large-scale-data-pipelines.)
The four mandated sections¶
| Section | Purpose | Why not "encyclopedia" |
|---|---|---|
| Quick Commands | Copy-paste ops the agent runs most | Not: prose describing the CLI surface area |
| Key Files | "The 3-5 files you actually need" | Not: a table of every file in the module |
| Non-Obvious Patterns | Tribal knowledge + gotchas | Not: a narrative of the module's design history |
| See Also | Cross-references to related modules + concepts | Not: full dependency graphs inline |
Why the shape matters¶
The principle is a token-budget discipline applied at the per-file level, composing with concepts/context-engineering at the per-turn level:
- ~1,000 tokens per file means loading 10 files still fits inside a generous headroom for tool outputs + conversation history.
- All 59 files together consume less than 0.1% of a modern model's context window (Meta's disclosure) — the entire knowledge layer fits inside the headroom of a single tool call.
- Four sections with explicit purposes means agents can section-hop rather than read linearly — "looking for the gotcha" is a direct jump.
- Short files mean fast maintenance — when the underlying code changes, 30 lines are easier to re-validate than 300.
Contrast with encyclopedia-style docs¶
Encyclopedic documentation (the academic-research comparison point: Django / matplotlib auto-generated context files) fails because:
- Context rot — long encyclopedic blobs dilute the signal the agent needs for this specific task.
- Pretraining overlap — on well-known OSS, the model already knows the encyclopedia; injecting it adds noise.
- Maintenance unfeasibility — 300-line files can't stay fresh at Meta's refactor pace.
2025 academic research found AI- generated context files decreased agent success rates on Django / matplotlib. Meta's response: "Files are concise (~1,000 tokens, not encyclopedic summaries), opt-in (loaded only when relevant, not always-on), and quality-gated."
Shared family with related disciplines¶
- patterns/tool-surface-minimization — same design posture at the tool-registry level: expose only the tools needed, not all tools.
- concepts/context-window-as-token-budget — the parent discipline compass-not-encyclopedia operates inside.
- concepts/context-rot — the failure mode compass-shape prevents.
- patterns/five-questions-knowledge-extraction — the extraction methodology that produces compass-shaped outputs: ask pointed questions, not "describe the module."
Applied more broadly¶
The principle generalises beyond code-navigation context files:
- LLM system prompts — the same discipline argues for task- focused 25-35-line system prompts over exhaustive role/persona pages.
- Runbooks for human + AI incident response — the "what specifically to do when alert X fires" shape beats the "full history of the service" shape.
- Migration guides — compass-shaped is "here are the 5 files you change and the 3 invariants you must not break"; encyclopedia- shaped is a 300-line narrative.
Seen in¶
- Meta AI Pre-Compute Engine (2026-04-06) — canonical wiki instance. 59 context files in compass-not-encyclopedia format cover 100% of modules in a 4,100-file config-as-code data pipeline. Combined size < 0.1% of a modern model's context window. Each file mandates Quick Commands / Key Files / Non-Obvious Patterns / See Also. (Source: sources/2026-04-06-meta-how-meta-used-ai-to-map-tribal-knowledge-in-large-scale-data-pipelines.)
Related¶
- concepts/tribal-knowledge — what goes in the Non-Obvious Patterns section
- concepts/context-engineering — the parent discipline
- concepts/context-window-as-token-budget — the cost axis compass-shape optimises
- concepts/context-file-freshness — the staleness discipline that compass-shape makes tractable
- patterns/precomputed-agent-context-files — the architectural pattern compass-not-encyclopedia is the format rule for
- patterns/tool-surface-minimization — the sibling discipline at the tool-registry level
- systems/meta-ai-precompute-engine — the canonical producer