briefd

[briefd logo]

briefd

Your agents, briefed. Not flooded.
A self-hosted context compiler for AI coding teams: git-backed knowledge, served to coding agents
as token-budgeted context bundles over MCP.

[CI] [Security] [Release] [Go version] [License] [Container] [MCP Registry] [Glama score] [M8ven Score]

[Today: CLAUDE.md is re-sent on every turn. With briefd: one compile_bundle call returns ~1,800 tokens. Result: 86% fewer knowledge tokens per task.]
▶ Watch the 90-second explainer · How it works · Quickstart

Teams that build many projects in one domain keep the same knowledge in their heads and in scattered CLAUDE.md / AGENTS.md files: terminology, business rules, architecture decisions, conventions. Loading all of it into every session burns thousands of tokens on every turn, and whatever doesn't fit gets left out.

briefd inverts the model: context on demand, not up front. Your knowledge lives as Markdown in a git repository. briefd indexes it and answers one question from your coding agent — "what do I need to know for this task?" — with a compiled, deduplicated bundle that never exceeds the token budget you set.

Why

On the sample knowledge repo in this repository (44 documents, 47 realistic developer tasks), compile_bundle spends 86% fewer tokens per task than pasting everything into CLAUDE.md while still containing the section that answers the task 96% of the time (98% with the English-only all-MiniLM-L6-v2 model, the figure the explainer video quotes). The realistic middle ground — a hand-curated CLAUDE.md with just conventions and the glossary — costs 2.8× more than a bundle and has the answer less than half the time.

Reproduce it with make bench; the method is in internal/eval/bench.go.

That is what the tokenizer says. Inside real Claude Code sessions (eval/session/, Sonnet, 10 tasks, same prompts) briefd cut the context carried per turn by 35% and the cost per task by 40% with identical answers — at the price of 3–4 extra tool-call round trips per task. The saving grows with the size of your knowledge repo; a static CLAUDE.md cannot.

Not another agent memory

Memory tools (agentmemory, Mem0, claude-mem) record what an agent observed in its sessions, automatically, per user. briefd serves what the team decided, written by people and reviewed like code. They answer different questions and run side by side.

Use a memory tool so your agent remembers what it tried last week. Use briefd so every agent on the team applies the same rules — and so someone notices when a rule falls behind the code.

How it works

• Git is the source of truth. The index is a disposable cache rebuilt from a clone.
• Agents never write to the index. propose_update opens a reviewable branch/PR; what briefd serves changes only when a human merges.
• Hybrid retrieval, no external services. SQLite FTS5 (BM25) + multilingual embeddings (multilingual-e5-small, 100+ languages) computed by a pure-Go encoder, fused with reciprocal rank fusion. No Postgres, no vector database, no ONNX runtime, no CGO.
• Hard token budgets. Every API that returns context takes max_tokens and never exceeds it.

Quickstart

Try it in 30 seconds — no repository needed:

briefd demo      # serves the built-in sample knowledge base on http://127.0.0.1:7788

Then open the dashboard, or point an agent at it: claude mcp add --transport http briefd http://127.0.0.1:7788/mcp. The first run downloads the embedding model (about 470 MB); briefd demo --embeddings none skips it.

With your own knowledge:

# 1. build (Go >= 1.26) or grab a binary from the releases pagegit clone https://github.com/ismailperim/briefd && cd briefd && make build
# 2. serve the sample knowledge repo (downloads the 470 MB multilingual embedding model once)./bin/briefd serve --source testdata/knowledge --db /tmp/briefd.db --token dev-token
# 3. connect Claude Codeclaude mcp add --transport http briefd http://localhost:7788/mcp \  --header "Authorization: Bearer dev-token"

Open http://localhost:7788/ for the dashboard, then ask Claude Code something the sample corpus knows — "what's our retry policy for acquirer calls?" or "ters ibraz nedir?" — and watch search_context / compile_bundle show up in the request log.

Any MCP client that speaks streamable HTTP works. For a project-level .mcp.json:

{  "mcpServers": {    "briefd": {      "type": "http",      "url": "http://localhost:7788/mcp",      "headers": { "Authorization": "Bearer dev-token" }    }  }}

Let the agent install it. Hand your coding agent one instruction:

Retrieve and follow the instructions at: https://raw.githubusercontent.com/ismailperim/briefd/main/INSTALL_FOR_AGENTS.md

Teach the agent when to ask. Tools an agent does not call save nothing. The skills/briefd skill tells Claude Code (and any agent that reads SKILL.md) when to compile a bundle, how to treat a stale section and when to propose an update; the same guidance as a CLAUDE.md paragraph is in deploy/local/CLAUDE.md.

npx skills add ismailperim/briefd

Single-user, no server? briefd mcp speaks MCP over stdio — the same tools, the same index, no port and no token. Claude Desktop, Cursor's stdio config and MCP directory inspectors launch it directly:

{  "mcpServers": {    "briefd": {      "command": "briefd",      "args": ["mcp", "--source", "/path/to/knowledge", "--db", "~/.briefd/knowledge.db"]    }  }}

Use serve when a team shares one instance (dashboard, metrics, webhook, REST); use mcp when the agent runs on the machine that holds the checkout.

Tools

The same operations are available over REST (/api/search, POST /api/bundle, /api/docs/{path}, /api/scopes, POST /api/proposals, POST /api/usage, /api/gaps, /api/health, /api/stats) behind the same bearer token.

Your knowledge repo

briefd expects a git repository (or directory) of Markdown with three kinds of folders (briefd init <dir> scaffolds it with example documents):

knowledge-repo/├── domain/          # shared: terminology, business rules, ADRs├── conventions/     # shared: coding standards, infra patterns└── projects/    ├── ledger-service/   # visible only when scope "projects/ledger-service" is requested    └── merchant-portal/

Documents are split on ##/### headings into sections of roughly 200–800 tokens with stable ids, so a section can be quoted on its own. Optional front matter adds metadata:

---title: Retry policy            # defaults to the first H1tags: [payments, resilience]refs: ["services/payment/**"]  # code paths this doc governs---

testdata/knowledge/ is a complete example (a fictional payments platform) and doubles as the evaluation corpus.

Obsidian vaults and links

A knowledge repository can be an Obsidian vault. briefd reads [[wikilinks]] in the text and in front-matter properties such as related: (including [[note|alias]] and [[note#heading]]) and relative Markdown links, resolves them the way Obsidian does (by path, or by file name anywhere in the repository), and builds a link graph: the dashboard draws it, get_document returns each document's links and backlinks so an agent can follow them, and orphans (nothing links here) and broken links (the target does not exist) are listed as maintenance signals next to gaps and coverage. Folder scopes still apply — domain/, conventions/, projects/<name>/ — so keep the vault's top level in that shape.

Running it for real

export BRIEFD_GIT_TOKEN=ghp_...     # only for private HTTPS remotes./bin/briefd serve --source https://github.com/your-org/knowledge.git --token "$(openssl rand -hex 16)"

briefd clones the repository, follows the branch with fetch + hard reset every sync.interval (default 60 s), or immediately when your forge calls POST /webhook/git with a GitHub-style HMAC signature. Only changed files are re-parsed and re-embedded.

Languages. The default embedding model, multilingual-e5-small, covers 100+ languages, so a Turkish, German or Japanese knowledge repo — or English docs queried in another language — works out of the box. English-only teams can set embeddings.model: all-MiniLM-L6-v2 (87 MB, ~2.5× faster indexing). briefd model pull pre-fetches a model for offline or image-build use; --embeddings none gives BM25-only mode; Ollama and OpenAI-compatible services are alternative providers.

Docker

cd deployBRIEFD_SOURCE=https://github.com/your-org/knowledge.git BRIEFD_API_TOKEN=... docker compose up

The image is distroless and pure Go (~34 MB, linux/amd64 + arm64). Database, checkout and model live in the briefd-data volume. Mount a directory and set BRIEFD_SOURCE=/knowledge to serve local files instead.

Deployment guide: deploy/README.md covers Compose and systemd setups, git forges (GitHub, GitLab, Azure DevOps, Bitbucket, SSH), installing the embedding model offline, proxies and private CAs, exposure/security, upgrades and monitoring. For a laptop-only setup see deploy/local/.

Configuration — briefd.yaml (see deploy/briefd.example.yaml) or BRIEFD_* environment variables; flags override both. The ones you will actually touch:

briefd model pull pre-fetches the embedding model for offline or image-build use.

Dashboard and metrics

GET / is a dashboard embedded in the binary (no build step, no external assets):

• Overview — requests, tokens served and saved, index and sync; the knowledge graph with the documents agents were served in the last 24 hours stamped on it; the circulation list; and Needs attention, one line per maintenance signal, worst first.
• Graph — every document and link, zoom and pan (mouse or keyboard), scope filters, search; select a document to see its links, backlinks and how often it was served (#graph=<doc path> links straight to it).
• Maintenance — proposals awaiting review, knowledge gaps, documents behind the code, the oldest documents, links to fix and code coverage.
• Activity — latency and volume per tool, the index by scope and the request log; counters persist across restarts and can be reset.
• Instance — the running configuration (secrets shown only as set/unset), with Sync now and Rebuild index.

GET /metrics exposes the counters in Prometheus text format; GET /api/stats as JSON.

Knowledge gaps. Every search_context / compile_bundle call is logged with its retrieval confidence (query_log, 30-day retention). The dashboard lists the questions of the last seven days that the knowledge base did not answer — nothing matched, or the agent's report_usage said no section helped — grouped by question and ranked by how often they were asked, plus the answered questions whose top result barely stood out from the rest. That list is the backlog for whoever maintains the repository; GET /api/gaps?days=7&limit=20 returns it as JSON.

Document age. Every section in a bundle carries the date its document last changed (## path — heading (updated 2026-03-04), from git history, or the file mtime for a plain directory), so an agent can weigh a rule by its age. The dashboard lists the documents that changed longest ago — the ones to re-read first.

Behind the code. Give a document refs: ["services/payment/**"] in its front matter and list the code repositories in code.repos; briefd follows their history (bare clones, never the files) and counts the commits that touched a governed path after the document last changed. The attribution line then reads (updated 2026-03-01; code changed since: 3 commits, last 2026-06-01), the dashboard lists the documents most behind, and briefd_documents_behind_code is exported. The agent reading a stale rule is often the right one to fix it with propose_update. Design in ADR-0007.

The same refs work the other way round: pass paths (the files the task touches) to compile_bundle and the rules that govern them lead the bundle, and the dashboard's Coverage panel lists the directories of each code repository that no document claims — the knowledge base's blind spots (ADR-0008). briefd does not index code itself; your agent's grep and LSP do that better.

Retrieval quality

Retrieval is measured, not assumed. make eval scores 47 English golden queries (keyword, paraphrase, typo, mixed-language) over the sample corpus and 30 Turkish queries over a Turkish corpus; CI fails if hybrid retrieval drops below eval/thresholds.yaml or eval/thresholds-tr.yaml:

On public BEIR datasets briefd's vector-only mode reproduces the published quality of both embedding models and hybrid mode beats BM25 and vector-only on each — SciFact nDCG@10 0.714 vs 0.665 for the BEIR BM25 baseline; see eval/beir/ to reproduce.

Every change to chunking, embeddings or fusion ships with before/after numbers (ADR-0004 is an example).

CLI

briefd demo       # serve the built-in sample knowledge base — try it in 30 secondsbriefd init       # scaffold a knowledge repo (domain/, conventions/, projects/)briefd serve      # MCP over HTTP + REST + dashboard, for a shared instancebriefd mcp        # MCP over stdio, for one agent on this machine (Claude Desktop, Cursor)briefd index      # index a directory into the database (--rebuild to start over)briefd search     # query like search_context does (--mode bm25|vector|hybrid, --json)briefd model list # local embedding models and whether they are downloadedbriefd model pull # download a model (--model all-MiniLM-L6-v2 for the English one)briefd eval       # retrieval quality against the golden setbriefd bench      # tokens per task: static CLAUDE.md vs compile_bundle

Status and roadmap

v0.1 is feature-complete; expect rough edges before 1.0. Planned next:

• usage-driven relevance tuning from report_usage
• contradiction detection for proposals
• a light Turkish stemmer for the BM25 side and glossary-alias query expansion
• multiple knowledge repositories per instance

Read docs/ARCHITECTURE.md for a guided tour with diagrams. The full specification is in SPEC.md; decisions are recorded in docs/adr/.

Contributing

Issues and pull requests are welcome — see CONTRIBUTING.md for the development setup, testing rules and conventions. Security issues: SECURITY.md.

License

Apache-2.0