Read CLAUDE.md before repository work. Binding routing: this repo owns skills, Engine, and MV3 extension; Agent SDK work belongs in /Users/julb/Desktop/GitHub/caveman-agent-sdk; Browse driver/MCP/benchmark/plugin work belongs in /Users/julb/Desktop/GitHub/caveman-browse; proprietary Pebble product work belongs in /Users/julb/Desktop/GitHub/caveman-coding-agent. Visibility: this repo and caveman-browse are public now; caveman-agent-sdk is private during development with planned public release after its gates; caveman-coding-agent and Caveman-Cloud remain private commercial source. @./skills/caveman/SKILL.md @./skills/caveman-commit/SKILL.md @./skills/caveman-review/SKILL.md @./skills/caveman-compress/SKILL.md
Steady7 mentions in 2 weeks
1 project uses a version of this
115k7mo agoReviewMountCreate a free account to mount
Repository routing: do not continue Browse product work here. Source of truth is JuliusBrussee/caveman-browse, local checkout /Users/julb/Desktop/GitHub/caveman-browse. This directory is a consumer copy; edit only for pinned integration, migration/removal, or an explicit cross-repo sync. Accessibility-compressor work stays here in engine/compressors/axtree.go because Engine is owned by this repo. Local, agent-facing browser interaction for the open Caveman surface. It serves stdio MCP tools that read a real Chrome accessibility tree, compress it with the engine's forced-only a11y compressor, act on uid handles,…
Steady7 mentions in 2 weeks
1 project uses a version of this
115k7mo agoReviewMountCreate a free account to mount
Detect a payload's type → route to a safety-classed compressor → count the token reduction → store the original for recovery. The stable four-call API (Compress/Retrieve/Detect/Stats) is what the proxy, CLI, SDKs, MCP, and WASM build all share. Everything it reports is inferred; it never says verified.
Steady7 mentions in 2 weeks
1 project uses a version of this
115k7mo agoReviewMountCreate a free account to mount
Detect a payload's type → route to a safety-classed compressor → count the token reduction → store the original for recovery. The stable four-call API (Compress/Retrieve/Detect/Stats) is what the proxy, CLI, SDKs, MCP, and WASM build all share. Everything it reports is inferred; it never says verified.
Steady7 mentions in 2 weeks
No other project uses it yet
115k7mo agoReviewMountCreate a free account to mount
The consumer top-of-funnel: a toggle that, on ChatGPT / Claude / Gemini, prepends the caveman skill directive to each message you send and re-fires the send, so the AI replies in a compact, high-density style intended to preserve technical substance. It only ever adds text to your outgoing message; it never reads replies, never touches network requests. Savings are inferred. (This is the directive-injector product; an earlier revision shipped a local WASM compress button — EXT removed it after Chrome…
Steady7 mentions in 2 weeks
No other project uses it yet
115k7mo agoReviewMountCreate a free account to mount
Declarative routing recipes for AI SDKs/frameworks: each recipe is the one-line base-URL shape that points a framework at the byte-safe gateway, with {{baseURL}} (gateway origin) and {{app}} (the /w/<app> attribution slug) left as render-time templates. Recipes are DATA, not code — the collapse decision applied to app frameworks. See ../../CLAUDE.md (root) · ../agents/CLAUDE.md (the agent twin) · ../cli/CLAUDE.md
Steady7 mentions in 2 weeks
1 project uses a version of this
115k7mo agoReviewMountCreate a free account to mount
A thin stdio JSON-RPC adapter exposing the compression engine as five MCP tools to any host (Claude Code, Cursor, …). It owns only the MCP framing; all compression is the engine's, linked in-process (no subprocess, no drift). Local-only — it opens no network connection — and everything it reports is inferred, never verified.
Steady7 mentions in 2 weeks
No other project uses it yet
115k7mo agoReviewMountCreate a free account to mount
A thin stdio JSON-RPC adapter exposing the compression engine as five MCP tools to any host (Claude Code, Cursor, …). It owns only the MCP framing; all compression is the engine's, linked in-process (no subprocess, no drift). Local-only — it opens no network connection — and everything it reports is inferred, never verified.
Steady7 mentions in 2 weeks
No other project uses it yet
115k7mo agoReviewMountCreate a free account to mount
Durable, cross-session memory: remember / recall / supersede / history / forget. A local SQLite store holds the raw memories (the source of truth); recall ranks them with deterministic BM25 behind a conservative threshold and compresses each hit through the engine so the inferred token cost is honest and the dropped detail stays recoverable via CCR. Everything is inferred. See ../../CLAUDE.md (root) · ../engine/CLAUDE.md · ../mcp/CLAUDE.md
Steady7 mentions in 2 weeks
No other project uses it yet
115k7mo agoReviewMountCreate a free account to mount
Durable, cross-session memory: remember / recall / supersede / history / forget. A local SQLite store holds the raw memories (the source of truth); recall ranks them with deterministic BM25 behind a conservative threshold and compresses each hit through the engine so the inferred token cost is honest and the dropped detail stays recoverable via CCR. Everything is inferred. See ../../CLAUDE.md (root) · ../engine/CLAUDE.md · ../mcp/CLAUDE.md
Steady7 mentions in 2 weeks
No other project uses it yet
115k7mo agoReviewMountCreate a free account to mount
A base-URL-swap reverse proxy: match → authenticate → inspect → byte-safe transform → upstream → meter. Single-operator, BYOK, zero cloud dependencies. It shares its provider adapters with the managed gateway (the managed gateway imports them from here). caveman start launches the caveman-proxy binary. - listener lifetime is not session lifetime: serve never exits because a wrapper, native session, heartbeat, or idle timer expires. Legacy CAVEMANNATIVEIDLE_TIMEOUT does not arm shutdown. The CLI never restarts a shared listener to change mode/recovery; an…
Steady7 mentions in 2 weeks
No other project uses it yet
115k7mo agoReviewMountCreate a free account to mount
A base-URL-swap reverse proxy: match → authenticate → inspect → byte-safe transform → upstream → meter. Single-operator, BYOK, zero cloud dependencies. It shares its provider adapters with the managed gateway (the managed gateway imports them from here). caveman start launches the caveman-proxy binary.
Steady7 mentions in 2 weeks
No other project uses it yet
115k7mo agoReviewMountCreate a free account to mount
Compress MCP/OpenAI tool definitions before they fill the context window. A thin Go wrapper over the engine's toolschema compressor: it drops annotation metadata (examples, titles, $comment, $schema) and reduces long descriptions to their lead plus every constraint-bearing sentence. It keeps every selection token (tool/parameter names, types, enums, required) and argument-construction value (default, const, $ref targets) byte-for-byte. Everything is inferred. Selection alone does not guarantee valid arguments. Short descriptions stay whole; long ones keep the lead plus every recognised constraint sentence…
Steady7 mentions in 2 weeks
No other project uses it yet
115k7mo agoReviewMountCreate a free account to mount
Compress MCP/OpenAI tool definitions before they fill the context window. A thin Go wrapper over the engine's toolschema compressor: it drops annotation metadata (examples, titles, $comment, $schema) and reduces free-text descriptions to their lead sentence plus every constraint-bearing sentence, while keeping every selection token — tool and parameter names, types, enums, required — and every argument-construction value — default, const, $ref targets — byte-for-byte. Everything is inferred.
Steady7 mentions in 2 weeks
No other project uses it yet
115k7mo agoReviewMountCreate a free account to mount
When to delegate to `cavecrew-investigator` (locate code), `cavecrew-builder` (1-2 file edit) or `cavecrew-reviewer` (diff review) instead of working inline or using `Explore`. Their output is compressed, so main context lasts longer.
Cooling7 mentions in 2 weeks
37 projects use a version of this
115k7mo agoReviewMountCreate a free account to mount
Find and label every LLM workflow in the repository so Caveman Cloud groups spend by workflow instead of one bucket. Use for "discover workflows" or breaking LLM spend down by workflow.
Steady7 mentions in 2 weeks
6 projects use a version of this
115k7mo agoReviewMountCreate a free account to mount
Read-only review of Caveman Cloud evidence: cost, Cave Score, workflows, traces, latency, errors, routing, savings. Use when asked what Caveman found or where LLM spend goes.
Steady7 mentions in 2 weeks
6 projects use a version of this
115k7mo agoReviewMountCreate a free account to mount
Read-only repository explorer for cold-start orientation, broad cross-file localization, or when a direct search failed. Skip it when the exact file or symbol is already named. Returns path:line citations only; its reads stay out of main context.
Surging7 mentions in 2 weeks
8 projects use a version of this
115k7mo agoReviewMountCreate a free account to mount
Plain text files in a repository that tell a coding agent how the project works: commands to run, conventions to follow and things to avoid. CLAUDE.md, AGENTS.md, cursor rules and skills are the common kinds.
CLAUDE.md or AGENTS.md?
CLAUDE.md is read by Claude Code. AGENTS.md is an open format that Codex, Cursor and other agents read. Many projects keep one and point the other at it.
What is a skill?
A folder with a SKILL.md that describes one capability, such as filling PDFs or reviewing code. The agent loads it only when the task calls for it.
Can I search my own team's files too?
Your agents already can, over MCP, limited to the files you're allowed to read. Searching them from this page is coming.
JuliusBrusseeAGENTS.md