Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/paulbreuler/limps/007-documentationgit clone --depth 1 https://github.com/paulbreuler/limpsWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00000 | $0.01837 |
| Opus 5 | $0.00000 | $0.00919 |
| Sonnet 5 | $0.00000 | $0.00367 |
| Haiku 4.5 | $0.00000 | $0.00184 |
Grade A, and why
007-documentation scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 301 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent 007: Documentation
Objective
Document the system-intelligent architecture, CLI reference, and integration with Plan 0028 (Obsidian).
Tasks
1. Architecture Overview (docs/knowledge-graph.md)
# Knowledge Graph Architecture
## Philosophy: System-Intelligent, Not AI-Intelligent
limps uses a knowledge graph to track relationships between plans, agents, features, and files.
The key principle: **the system is smart, not the AI**.
### What This Means
| AI-Dependent (Wrong) | System-Intelligent (Right) |
|---------------------|---------------------------|
| AI decides which tool to call | System routes deterministically |
| AI reasons about retrieval | Regex patterns select strategy |
| AI asks "are there conflicts?" | System surfaces conflicts proactively |
| LLM extracts entities | Regex + lightweight NLP extracts |
### Architecture
File changes → Auto-index → Conflict detection → Notify (no query needed)
User query → Deterministic router → Hybrid retrieval → AI formats output
### Interface Hierarchy
CLI (source of truth) ├── MCP (wrapper for AI that can't shell out) └── Obsidian Plugin (human UI, wraps CLI)
## Entity Types
- **plan**: A feature plan (e.g., `plan:0042`)
- **agent**: A task within a plan (e.g., `agent:0042#003`)
- **feature**: A feature within a plan (e.g., `feature:0042#1`)
- **file**: A source file (e.g., `file:src/auth.ts`)
- **tag**: A tag (e.g., `tag:knowledge-graph`)
- **concept**: An extracted concept (e.g., `concept:authentication`)
## Relationship Types
- **CONTAINS**: plan → agent, plan → feature
- **DEPENDS_ON**: agent → agent
- **MODIFIES**: agent → file
- **IMPLEMENTS**: agent → feature
- **SIMILAR_TO**: feature → feature (with confidence score)
- **BLOCKS**: derived inverse of DEPENDS_ON
## Conflict Detection
| Conflict | Severity | Description |
|----------|----------|-------------|
| File contention (2 WIP) | Critical | Two WIP agents modify same file |
| File contention (WIP + GAP) | Warning | WIP and GAP agents will modify same file |
| Feature overlap (>85%) | Warning | Features are very similar |
| Circular dependency | Critical | A → B → C → A |
| Stale WIP (>14 days) | Critical | Agent stuck in WIP |
| Stale WIP (>7 days) | Warning | Agent possibly stuck |
## Hybrid Retrieval
Queries are routed deterministically based on patterns:
| Pattern | Strategy | Example |
|---------|----------|---------|
| `plan \d+`, `agent #\d+` | Lexical first | "plan 0042" |
| `depends`, `blocks`, `modifies` | Graph first | "what blocks 003" |
| `how`, `why`, `explain` | Semantic first | "explain auth flow" |
| `status`, `progress` | Graph + Lexical | "status of plan 41" |
Results are fused using Reciprocal Rank Fusion (RRF).
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 301 lines · 0 tokens per session scan A d984d2601823
007-documentation is an agent published in the GitHub repository paulbreuler/limps (10 stars, last pushed 6mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,837 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other agents, from other repositories
monitor
Reviews code for correctness, standards, security, and testability (MAP).
analyzer
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
comparator
Compare two outputs WITHOUT knowing which skill version produced them.
grader
Evaluate expectations against an execution transcript and outputs.
planner
Autonomous plan quality reviewer using Universal Planning Framework.
polymath
Cross-disciplinary synthesis; spawns domain-specific subagents, synthesizes findings across domains, and produces integrated insights.