CapyPilot

English | 简体中文

Your next chapter, one confident step at a time.

A local-first career workspace: discover opportunities, prepare truthful applications, and keep a clear record of every outcome.

CapyPilot helps one job seeker discover verifiable openings, compare fit, prepare truthful materials, complete supported application forms under explicit authorization, and confirm what was actually submitted. Profiles, resumes, credentials, browser sessions, receipts, and runtime logs stay on the user's machine by default.

CapyPilot is built for supervised execution, not blind bulk submission. A clicked Submit button is not counted as success; only decisive evidence matched to the exact job can create a durable submitted record.

Current status

• Beta, local-first, and CLI-first. The product runs on the user's machine; there is no hosted CapyPilot service, account system, or cloud sync.
• Useful end to end today. The current code supports official-source discovery and manual imports, fit scoring, validated resume reuse, truthful tailoring, cover-letter and PDF preparation, authorized browser assistance, application history, and receipt reconciliation.
• The browser dashboard is read-only. It presents the same four stages— Discover, Decide, Prepare, and Verify—without mutating the database or executing commands.
• Human review remains part of the product. Unsupported material answers, CAPTCHA or MFA, assessments, identity or financial documents, account recovery, security changes, and uncertain submission outcomes stop for review.
• CapyPilot v0.5.1 introduces our independent product brand, with a capybara identity and a calm, deliberate approach to career progress. Existing commands and local data remain compatible.

Product workflow

Stage	What CapyPilot does	What it does not assume
Discover	Collects official openings, optional board results, and manual leads with source state	A lead is not automatically a verified job
Decide	Checks eligibility, enriches descriptions, scores fit, and records readiness	A high score is not authorization to apply
Prepare	Routes validated resumes, identifies evidence gaps, tailors content, and validates PDFs	Missing facts are not invented
Verify	Separates authorization, browser observations, platform state, and durable receipts	A preview or final click is not proof of acceptance

The safe application path is:

prepare -> audit -> authorize -> submit -> observe -> reconcile receipt

If acceptance cannot be proved, the application remains submission_uncertain and is not submitted again automatically.

Install

Python 3.11 or 3.12 is recommended for the full workflow. Core commands and the official-source radar also support Python 3.13.

There is currently no applypilot-local release on PyPI. Install from the latest GitHub release or directly from the repository:

pipx install "git+https://github.com/raederhans/ApplyPilot.git@v0.5.1"

For a reproducible local deployment, download the v0.5.1 bundle and SHA256SUMS from the release page, verify the archive checksum, then extract the bundle and run the guided installer:

python install.py

Third-party job-board discovery is optional and currently intended for Python 3.11–3.12:

python install.py --with-jobboards

Quick start

Initialize the local workspace and check available capabilities:

applypilot init
applypilot doctor
applypilot dashboard

Discover and prepare opportunities:

applypilot radar collect
applypilot radar report --hours 24
applypilot run discover enrich score tailor cover pdf

Review one exact job before any submission:

applypilot review-readiness
applypilot apply --dry-run --url <verified-job-url>
applypilot authorize-batch --url <verified-job-url>
applypilot apply --authorization-file <batch-manifest.json>
applypilot reconcile-receipts --file <receipt.json>

Validated resume variants can be inspected and routed independently:

applypilot resume-library-sync
applypilot resume-library-status
applypilot resume-route --url <verified-job-url>

Run applypilot --help for the complete command list. Optional browser backends, interaction modes, provider behavior, and operating details belong in the command help and product documentation rather than this overview.

Compatibility and local data

The public product name is CapyPilot, while these technical identifiers remain unchanged during the migration:

• distribution: applypilot-local
• Python package and CLI: applypilot
• environment variables: APPLYPILOT_*
• default workspace: ~/.applypilot/
• database schema, storage keys, entry points, and repository URL

Do not commit or share the local workspace. It may contain profile data, resumes, generated documents, SQLite databases, API keys, browser profiles, screenshots, receipts, logs, or verification codes. CapyPilot does not endorse CAPTCHA bypass, hidden submission, identity-document automation, or account recovery automation. See SECURITY.md before reporting a vulnerability.

Development and documentation

python -m pip install -e ".[dev]"
ruff check src
pytest -q
python scripts/build_release.py

• Product and frontend boundaries
• Changelog
• Contributing
• Security policy
• License and provenance

CapyPilot is licensed under GNU AGPL-3.0-only. It is an independent continuation of Pickle-Pixel/ApplyPilot; the original authors retain copyright in the upstream work. This repository is not affiliated with applypilot.app, useapplypilot.com, or similarly named products.