CtrlK
BlogDocsLog inGet started
Tessl Logo

pantheon-ai/implementation-planner

Converts a PRD or requirements document into a structured, phased implementation plan with individual phase files and granular per-task files written to .context/plans/. Also restructures existing monolithic planning documents into digestible, hierarchical directory structures. Creates a root plan index summarising all phases, a numbered phase file per phase, and a numbered task file per task inside each phase directory.

92

3.25x
Quality

93%

Does it follow best practices?

Impact

91%

3.25x

Average score across 5 eval scenarios

SecuritybySnyk

Passed

No known issues

Overview
Quality
Evals
Security
Files

criteria.jsonevals/scenario-1/

{
  "context": "Agent was asked to create a new implementation plan from a URL Shortener PRD. Evaluate the files produced under .context/plans/.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "plan-root-created",
      "description": "A plan root directory exists at .context/plans/plan-<slug>/ with a README.md file",
      "max_score": 5
    },
    {
      "name": "root-readme-is-index-only",
      "description": "The root README.md acts as a navigation index — it lists phases and links to them but does not contain implementation details, code, or per-task instructions",
      "max_score": 6
    },
    {
      "name": "phases-directory-structure",
      "description": "A phases/ subdirectory exists containing at least one numbered phase directory (e.g. phases/phase-01-<slug>/)",
      "max_score": 5
    },
    {
      "name": "phase-readme-has-goal-gate-deps",
      "description": "Every phase README.md contains a clear goal statement, a pass/fail gate that can be verified by running a shell command, and explicit dependencies on prior phases or artefacts",
      "max_score": 10
    },
    {
      "name": "gate-is-concrete-not-vague",
      "description": "No phase gate uses vague language like 'works correctly' or 'tests pass' without specifying the exact command to run",
      "max_score": 6
    },
    {
      "name": "tasks-directory-exists-per-phase",
      "description": "Each phase directory contains a tasks/ subdirectory with at least one task file",
      "max_score": 4
    },
    {
      "name": "task-identifier-format-correct",
      "description": "All task filenames follow the pattern task-P{NN}T{NN}-<slug>.md with zero-padded, 1-based numbering (e.g. task-P01T01-*.md, task-P02T03-*.md)",
      "max_score": 8
    },
    {
      "name": "task-files-have-verification-section",
      "description": "Every task file includes a verification section with a provable, runnable check (a shell command that exits 0/non-zero, file exists check, HTTP status check, etc.)",
      "max_score": 10
    },
    {
      "name": "task-verification-not-vague",
      "description": "No task verification section uses vague language — each verification describes a specific command or observable output, not 'the feature works'",
      "max_score": 8
    },
    {
      "name": "tasks-scoped-to-single-file",
      "description": "Task files are scoped to a single file or a tightly coupled set of files; no task spans multiple unrelated files without a declared reason",
      "max_score": 7
    },
    {
      "name": "slug-format-correct",
      "description": "All directory names and file slugs use lowercase kebab-case (e.g. plan-url-shortener-service, not plan-URLShortener or plan_url_shortener)",
      "max_score": 4
    },
    {
      "name": "scripts-used-for-scaffolding",
      "description": "The agent invoked new-plan.sh, new-phase.sh, and new-task.sh scripts to scaffold the directory structure rather than creating directories manually",
      "max_score": 8
    },
    {
      "name": "validate-plan-script-run",
      "description": "The agent ran validate-plan.sh <plan-slug> after writing all files and addressed any reported failures before completing",
      "max_score": 8
    },
    {
      "name": "completion-summary-reported",
      "description": "Agent output a structured completion summary listing the root README.md, each phase README with its task count, and all individual task file paths",
      "max_score": 4
    },
    {
      "name": "additive-no-deletion",
      "description": "The agent did not delete or overwrite any pre-existing files in .context/plans/ (files are additive)",
      "max_score": 3
    },
    {
      "name": "reasonable-phase-count",
      "description": "The plan contains between 2 and 8 phases appropriate for the project scope; if >8 phases were needed the agent asked the user before proceeding",
      "max_score": 4
    }
  ]
}

evals

scenario-1

criteria.json

task.md

SKILL.md

tile.json