perch-agent
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| PERCH_AGENT_CLI | No | Explicit path to the CLI's JS entry (auto-resolved by default). Point it at the `.js`, not the `.cmd`. | |
| PERCH_AGENT_MODEL | No | Default model pin for every delegation. | standard |
| PERCH_AGENT_EFFORT | No | Default reasoning effort. | high |
| PERCH_AGENT_DEFAULT_CWD | No | Fallback `cwd` when the caller omits one. Defaults to the OS temp dir. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| delegate_to_perchA | Delegate a task to Perch AI (worker agent) via the official perch CLI. Perch runs autonomously in the given working directory with its own tools (bash, file read/write, skills) on the signed-in Perch plan. Karpathy working guidelines (assumptions, simplicity, surgical scope, evidence-based verification) are prepended to every task by default (guidelines:false to disable). Returns the final answer, thread id, and status. Long-running tasks: pass wait_seconds to bound blocking; if still running you get a job_id — poll perch_job_status. |
| perch_job_statusC | Get status/result of a delegated Perch job by job_id. |
| perch_job_cancelB | Kill a running delegated Perch job (process tree). |
| perch_statusA | Free local check (no model call): perch CLI presence and sign-in state, plus this server's live job list. Call before delegating if unsure the worker is ready. |
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 4 tools
Each tool maps to a distinct phase of the delegation lifecycle: delegate, poll a job, cancel a job, and check local CLI readiness. The only mild overlap is between perch_status and perch_job_status, but descriptions make clear one is a per-job query and the other a free local sign-in/health check.
All names are snake_case and share a perch_ namespace, but the verb/noun ordering is mixed: delegate_to_perch is verb-first, while perch_job_status and perch_job_cancel are noun-first. It is still readable but not a single predictable pattern.
Four tools is thin but well-matched to a focused delegation wrapper: one action plus the poll/cancel/health trio needed to operate it. Nothing feels redundant or gratuitous.
The delegated-job lifecycle is covered end to end: readiness check, launch with optional bounded wait, status polling, and cancellation. Minor gaps like explicit job listing or cleanup are largely absorbed by perch_status reporting the live job list.