caprock
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
Caprock records every coding-agent session on this computer. Use these read-only tools to recall earlier work: search_memories, current_decisions and next_session_handoff for decisions and findings kept in Second Brain (search returns ranked candidates to judge by meaning; rephrase and retry when none fits), recent_sessions, session_summary and session_prose for what sessions in this repository did and wrote, search_prose to find a past explanation, cost for spend and tokens, ci_status for GitHub Actions, loop_alerts for sessions stuck repeating a tool call. Memories and prose may be stale: check them against the code before relying on them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| search_memoriesA | Search the user's Second Brain: decisions, findings and notes kept from earlier coding sessions, across every project or one. Returns up to 30 ranked candidates (any word matches; rarer words, and hits in a title or its English keywords, rank higher; a few synonyms such as PR/pull request, cost/price/spend and decision/rule are matched; typos are tolerated), each with kind, date, project, keywords and a short summary: judge by meaning which ones answer the question, then read one with get_memory. Titles are often in the user's own language, not English, while keywords are English. If nothing fits, rephrase and search again: other words, fewer words, a synonym, the product or component name, or words in the user's language. A memory marked superseded_by is no longer current: the id named replaces it. |
| current_decisionsA | The decisions in force for a project, newest first: every decision in Second Brain that no newer one supersedes. Read this before changing direction on something the user already decided. |
| get_memoryA | One Second Brain memory in full, with what it is linked to (topics, people, projects, related memories). |
| memory_topicsB | The topics Second Brain has memories about, most-linked first, and how many memories each project holds. |
| next_session_handoffB | What Caprock hands a new session started in a repository: the Second Brain memories it carries, whether the handoff is switched on, and when the passage it would quote was written. |
| recent_sessionsA | Recent coding-agent sessions (Claude Code, Codex, OpenCode, Gemini and others) in a repository, newest first: title, status, branch, model, what it is doing, cost and tokens, and a loop alert when one is active. |
| session_summaryA | One session: its title and description, status, branch, cost and tokens, turns and tool calls, the files it touched and the pull requests it opened. |
| session_proseA | The prose the agent wrote in a session (its answers and explanations, not tool output), newest first. Notes marked fragment read as mid-thought. |
| search_proseA | Search the prose agents wrote across every session on this computer — "which session explained the SSO change?" Newest first. |
| costA | Agent spend and tokens per day for a repository (or all of them), with totals by model. Figures are Caprock's, priced from its model table. |
| ci_statusA | GitHub Actions for a repository: the newest run of each workflow on its default branch, and the checks of the newest session's open pull request, with the failing jobs named. Needs GitHub connected in Caprock's Settings; read through Caprock, which already holds that connection. |
| loop_alertsB | Running sessions that are repeating the same tool call — a likely loop — with the tool, how many times, the repeated input and what the repeats paid to re-read the conversation. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 12 tools
Most tools target clearly distinct resources: sessions (recent_sessions, session_summary, session_prose), memories (search_memories, get_memory, current_decisions, memory_topics), and diagnostics (loop_alerts, cost, ci_status). The only confusable pairs are search_prose vs session_prose (cross-session search vs one-session read) and recent_sessions vs session_summary (list vs detail), but the descriptions explicitly differentiate both.
Two conventions coexist: noun-phrase names (ci_status, cost, loop_alerts, recent_sessions, session_summary, memory_topics, current_decisions, next_session_handoff) and verb_noun names (search_prose, search_memories, get_memory). The verb tools are internally consistent, but the overall set mixes styles, and bare `cost` is noticeably vaguer than its siblings.
Twelve tools is squarely in the well-scoped range, and each one maps to a distinct question an agent would ask about sessions, spend, CI, or stored memories. Nothing appears redundant enough to cut.
The read surface is broad: session lifecycle (list, detail, prose, loops, cost, CI, handoff) and memory retrieval (search, full read, topics, in-force decisions) are both covered. Gaps are minor and likely intentional for an observability server — no way to write a memory, record a decision, or enumerate projects directly.