Skip to content

v0.1.0

Choose a tag to compare

View all tags
@KeelDev-tech KeelDev-tech released this 16 Sep 01:33
· 56 commits to main since this release
v0.1.0
e06731a

v0.1.0 — First public release

Keel — honest automation, end to end. An open-core job-application autopilot with a truthfulness contract: it only ever claims what you tell it is true — verified facts, a canonical answer bank, banded rules for gray areas, and hard gates that stop the run when a question can't be answered honestly.

Three rules run the system: (1) truthfulness gates — never invents experience, degrees, or answers; (2) explicit confirmation — nothing counts as submitted unless the page itself confirms it; (3) fail closed — ambiguity parks the lead and moves on, never guesses.

What's in this release

  • Discovery/scoring engine — search playbooks, sweep prompts, and the 100-point fit model with a generic scorer template.
  • Truthful resume tailoring — templates driven by the operator's applicant profile, with truthfulness gates.
  • Sanitized answer bank — canonical answers plus banded-question rules and hard gates (ships with example data only — real profiles are never committed).
  • Prescreen gates — automated checks that park leads needing operator input instead of inventing answers.
  • ATS detection and capability radar — platform identification plus HTTP-level probes of which submission paths are viable (detection only — no submission behavior).
  • Launch-packet builder — the public application loop prepares complete launch packets and stops there, per the open-core boundary in SPLIT.md.
  • verify_retry — verification worker that re-checks parked leads over plain HTTP and promotes live postings or marks dead ones.
  • Feeder watchdog — watches the ready queue and verification pool so a dry queue surfaces instead of going silent.
  • Telemetry framework — append-only, additive-only event logging.
  • Outcome analytics — ledger + telemetry analysis with fail-closed reporting.
  • Pluggable inbox listener — template for wiring operator responses back into the system.
  • HTML dashboard builder — self-contained dashboard generated from the ledger and queues.
  • Setup wizard and stock config — guided first run with sane defaults and example data.

Open-core on purpose (Apache-2.0). Public: discovery/scoring, truthful resume tailoring, answer bank, prescreen gates, ATS detection + capability radar, launch-packet builder, telemetry, dashboard. Private by design: submission-behavior techniques ATS vendors could fingerprint — publishing them would get everyone's pipeline blocked.

Get started: from the source zip below (or a clone), run ./setup.sh, edit data/applicant_profile.json and data/answer_bank.json with your truth, then ./start.sh. Python 3.10+ stdlib only — no dependencies to install.

Full changelog: CHANGELOG.md. Public/private boundary: SPLIT.md. Repo: https://github.com/KeelDev-tech/keel

Assets 2
Loading