Skip to main content
Glama
Sign In
Keshro MCP

Keshro MCP

by jlewitt1
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

Keshro MCP

The intelligent execution layer for coding agents, exposed as an MCP server for high-stakes engineering projects.

pip install keshro-mcp

When to use this vs the CLI

Use the CLI (pip install keshro) for the full experience: interactive clarifying questions, migration detection, parallel execution in isolated worktrees, git checkpoints, cross-task context routing, and cost tracking.

Use MCP if your agent platform speaks MCP and you want direct tool-call access to Keshro plans and tasks.

The CLI gives you more control. MCP is more flexible for custom integrations.

Related MCP server: projscan

Setup

Set your API token:

export KESHRO_API_TOKEN="ksh_pat_..."

Get one from keshro.com/account.

Connect to your agent

MCP works with any agent that supports the protocol — Claude Code, Cline, Continue, Zed, and others.

Claude Code — add to ~/.claude.json:

{
  "mcpServers": {
    "keshro": {
      "command": "keshro-mcp",
      "env": { "KESHRO_API_TOKEN": "ksh_pat_..." }
    }
  }
}

Other MCP clients — point your client at the keshro-mcp binary with KESHRO_API_TOKEN set in the environment. The server uses stdio transport.

Available tools

Tool

What it does

preview_plan

Run Keshro's pre-plan intake and clarifying-question preview

generate_plan

Generate a plan from a description using AI

list_plans

List all plans

get_plan

Get a plan with all tasks

plan_status

Progress summary (task counts, enrichment sources)

next_task

Get the next actionable task

create_plan

Create a plan manually

start_task

Mark a task as in progress

complete_task

Mark a task as done

block_task

Mark a task as blocked

unblock_task

Clear a blocker

append_task_note

Add a note to a task

add_task_artifact

Attach an artifact link

record_decision

Log a decision with context, choice, and reasoning

edit_task

Edit task title or description

push_to_tracker

Push tasks to Linear, Jira, or GitHub as issues

sync_pull

Pull status updates from connected issue tracker

export_project

Export project data

Current parity notes

MCP now supports the newer task controls exposed in the web product:

  • explicit depends_on task dependencies

  • parallelizable task scheduling hints

  • per-task executor selection

  • generic issue linking via issue_id, plus external issue fields

  • pre-plan intake via preview_plan

It still remains thinner than the CLI for actual execution orchestration. The CLI owns parallel local worktrees, git checkpoints, richer execution transcripts, and the direct keshro continue runtime loop.

License

MIT

Releases

Publish the MCP package with one GitHub Actions run after you bump pyproject.toml:

gh workflow run "Publish MCP"

That workflow reads the package version from pyproject.toml, publishes the package to PyPI, then creates the matching vX.Y.Z GitHub release automatically.

Available Tools

25 tools
add_taskD

Inputs: plan_id (required), title (required), description (required), status, +12 more. Returns: object

add_task_artifactD

Inputs: plan_id (required), task_id (required), artifact_link (required), feedback_reason. Returns: object

append_replan_noteD

Inputs: plan_id (required), note (required). Returns: object

append_task_noteD

Inputs: plan_id (required), task_id (required), note (required), feedback_reason. Returns: object

block_taskD

Inputs: plan_id (required), task_id (required), blocked_reason (required), feedback_reason. Returns: object

complete_taskD

Inputs: plan_id (required), task_id (required), notes, feedback_reason, +1 more. Returns: object

create_planD

Inputs: migration_id (required), template_key, title, summary, +2 more. Returns: object

edit_taskD

Inputs: plan_id (required), task_id (required), title, description, +13 more. Returns: object

export_projectD

Inputs: migration_id (required). Returns: result

generate_planD

Inputs: description (required), title, project_type, repo, +1 more. Returns: object

get_historyD

Inputs: migration_id, plan_id. Returns: object

get_planD

Inputs: migration_id, plan_id. Returns: object

get_projectD

Inputs: migration_id (required). Returns: object

list_plansD

Inputs: search, limit, org_id. Returns: result

list_projectsD

Inputs: search, limit, org_id, status. Returns: result

list_templatesD

Inputs: template_key. Returns: result

next_taskD

Inputs: plan_id (required). Returns: object

plan_statusD

Inputs: plan_id (required). Returns: object

preview_planD

Inputs: description (required), title, project_type, repo, +1 more. Returns: object

push_to_trackerD

Inputs: plan_id (required), provider, team_id, project_id. Returns: object

record_decisionD

Inputs: plan_id (required), task_id (required), context (required), choice (required), +2 more. Returns: object

start_taskD

Inputs: plan_id (required), task_id (required), owner, owner_user_id, +3 more. Returns: object

sync_pullD

Inputs: plan_id (required). Returns: object

unblock_taskD

Inputs: plan_id (required), task_id (required), notes, feedback_reason, +1 more. Returns: object

update_planD

Inputs: plan_id (required), title, summary, status. Returns: object

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Deploy Server
    One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

    TDQS

    D1.8/5.0

    Scored across 25 tools

    Disambiguation4/5

    Tools are mostly distinct, with clear verbs like add, block, complete, start, unblock. However, 'create_plan' and 'generate_plan' may cause confusion, as well as 'plan_status' vs 'get_plan', though context may differentiate them.

    Naming Consistency5/5

    All tool names follow a consistent verb_noun pattern using snake_case (e.g., add_task, get_plan, list_projects). No mixing of conventions.

    Tool Count3/5

    25 tools is on the high side, exceeding the typical well-scoped range of 3-15. While the domain is broad, the count feels slightly heavy and may include some redundancy.

    Completeness3/5

    The tool set covers task, plan, and project lifecycles well, but lacks delete operations for any resource, which is a notable gap. Update is present only for plans, not projects.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Resources

    Related MCP Connectors

    Related MCP Servers

    • MCP Plus

      David3u
      F
      license
      Not graded
      quality
      D
      maintenance
      A lightweight MCP server that enhances AI agents with tools for codebase analysis, task delegation to sub-agents, multi-agent coordination through chatrooms, and project todo management.
      -

    • projscan

      abhiyoheswaran1
      A
      license
      Not graded
      quality
      C
      maintenance
      An MCP server that provides AI coding agents with AST-accurate, context-budget-aware codebase querying, safety gates, and team policy integration via structured tools and a local plugin layer.
      112 npm
      4
      MIT

    • knitbrain

      PDgit12
      A
      license
      A
      quality
      C
      maintenance
      Local-first MCP server that gives any AI coding agent per-project memory, workflow intelligence, and always-on, lossless token & context optimization.
      37
      65 npm
      6
      MIT