Commit Graph
157 Commits
Author SHA1 Message Date
WolfandClaude Fable 5.1 b64d77d224 Drop the waits on the landed remote and review plans and the mentions of the result files that never existed from the hosts plan
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-10-05 09:30:15 +02:00
WolfandClaude Fable 5.1 1b24de6980 Keep only the per-host filter parameters in the filter survey and name all twelve readers that declare a filter
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-10-05 09:30:01 +02:00
WolfandClaude Opus 5.5 1127d06f03 Record the judge scale, summarise on both orders and hidden companies in decisions.md
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 21:43:10 +02:00
Wolf 97c3046f53 architecture doc 2026-10-04 20:44:26 +02:00
WolfandClaude Opus 5.5 f00fdda47d Make the architecture document standalone: four-part fetch gate, closing, copies, defaults, job likes, summarise on both feed scores
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 20:39:38 +02:00
Wolf c98f8dcf28 architecture doc 2026-10-04 20:36:26 +02:00
WolfandClaude Opus 5.5 f7917a6fad Delete the ranking flow plan: its four packages are merged and its reasoning is in decisions.md
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 15:37:18 +02:00
WolfandClaude Opus 5.5 93e5e9f647 Bring the open items to fetch text and drop the starred-company one, since a star no longer widens
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 15:23:29 +02:00
WolfandClaude Opus 5.5 47e69df388 Fold the ranking flow decisions into decisions.md: three decisions per agent, the cut, fetch text, terms on every big board, favourite and the company switches, saved filters
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 15:23:01 +02:00
WolfandClaude Opus 5.5 060ccf66e6 Cut the ranking flow plan into four work packages with a contract
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 14:41:31 +02:00
WolfandClaude Opus 5.5 6741ce88a1 Put the fetch gate inside fetch as region, place and role, with All countries and All roles as separate overrides
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 14:32:43 +02:00
WolfandClaude Opus 5.5 4565507d7a Write the target architecture on its own and keep today versus decided as the ranking flow plan
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 14:22:19 +02:00
WolfandClaude Opus 5.5 7197b993c5 Settle the final score as fit, newness, affinity and likes, with everything else left to filters
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 14:20:18 +02:00
WolfandClaude Opus 5.5 82e4b928ee Merge All countries and All roles into one All roles switch that passes a whole company through the fetch gate; favourite only raises the level
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 10:49:23 +02:00
WolfandClaude Opus 5.5 28c13beda5 Frame the flow as three scores: fetch gate, analysis score, final score; the feed shows everything the fetch gate passes, with saved filters
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 10:38:59 +02:00
WolfandClaude Opus 5.5 672c9bab25 Keep a count and a minimum level on each list of the cut
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 10:24:01 +02:00
WolfandClaude Opus 5.5 e9a24815f7 Settle that a liked or saved role always makes the cut
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 10:18:49 +02:00
WolfandClaude Opus 5.5 1e039230ec Keep hidden as the only way down for a company
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 10:16:44 +02:00
WolfandClaude Opus 5.5 c9f401eb27 Settle company affinity as a bonus on the level of both lists in the cut
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 10:15:32 +02:00
WolfandClaude Opus 5.5 5f1d225d21 Accept the full re-judge after a CV or prompt edit
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 10:13:10 +02:00
WolfandClaude Opus 5.5 9d30582599 Make the flow the owner's architecture document, and move summarise after the ranking on the top of the feed
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 10:10:38 +02:00
WolfandClaude Opus 5.5 ccf6b3fd1c Settle that extract and judge stay two calls
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 08:54:43 +02:00
WolfandClaude Opus 5.5 16e0323198 Settle one depth for the model pass: fetch text, extract, rescore and judge read the whole cut, summarise follows
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 08:54:30 +02:00
WolfandClaude Opus 5.5 52d78960ce Settle the analysis cut as two joined lists, by prompt fit and by CV closeness
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 08:44:58 +02:00
WolfandClaude Opus 5.5 9be3e34020 Settle that location guards the analysis cut and that term-narrowed boards close by age only
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 08:42:49 +02:00
WolfandClaude Opus 5.5 3012afdd57 Write the ranking flow as one shared picture: fetch, cheap pass, second fetch, model pass, feed, with what is decided, proposed and open
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 08:23:38 +02:00
WolfandClaude Opus 5.5 fb24c8e4c9 Lead a role in several cities with the one nearest the viewing agent, in the row, the reader and the posting link
Places are ordered in tiers off the agent's gates (the any group, then the require scope, then the rest), and within a tier by the home the any group gives, since the gazetteer holds no coordinates. GET /api/roles and /api/roles/{uid} carry place_texts, lead_place and lead_url for the roles served; the gates and the dupes rule are unchanged.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-04 08:15:26 +02:00
WolfandClaude Opus 5.5 c23594507a Score each role's closeness to the agent's CV, let the judge also read the 50 closest the score leaves out, and order CV match's unjudged roles by it
The CV is cut into chunks inside e5's window, embedded like a prompt and
cached per agent under a hash of the model, chunking and text. A role's
cv_fit (scores.cv_fit, migration 31) is its best chunk cosine, scaled with
its own calibration band: fit's clipped 88 of agent 2's ungated roles to 1.
It is no term of the Agent match score. judge_cv_top (default 50) sits
beside judge_top and judge_min on the agents page and card.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-03 23:35:21 +02:00
WolfandClaude Opus 5.5 b481cb5692 Merge summaries for saved and liked roles and the order buttons Agent match, CV match, Newest (wp-summaries)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-03 23:14:30 +02:00
WolfandClaude Sonnet 5.5 ee8e450640 Summarise every saved or liked role whatever its rank, queue it at once on save or like, and shorten the order labels to Agent match, CV match and Newest so they fit one line
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-03 23:14:12 +02:00
WolfandClaude Opus 5.5 c7e1921be8 Pin that a Europe gate reads a town by its own country and a kept role through its copies' places
The OpenAI and Mistral roles agent 2 let through are not a gazetteer mix-up:
Montréal resolves to Canada and San Francisco to the US. Each is the kept
role of a posting folded with a copy in Dublin, Munich or Paris, and the
gate reads every place of a kept role. Tests hold both, decisions.md says so.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-03 23:13:05 +02:00
WolfandClaude Sonnet 5.5 8c6cbdf002 Give an agent a soft profile of likes and dislikes and take the must-haves off the form (wp-likes)
Add agents.like_text (migration 30) beside avoid_text, now shown as "What I don't like". The judge reads both under <likes> and <dislikes>, its career rubric judges against the prompts, the likes and the dislikes, and FORMAT 3 makes earlier judgements be made again. The form drops the must-haves box and no longer sends must_text, so a stored one is kept and still caps career in code.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-03 22:48:46 +02:00
WolfandClaude Sonnet 5.5 001fc9002b Let the model report a verdict per must-have and cap career at 39 in code when one is broken
The judge numbers an agent's must-haves (one per line) and answers met, broken or unclear for each; career is stored as the model gave it and counts as at most BROKEN_CAP when a role is served, in cv_score, Store.judgement and the SQL of Best CV match. Migration 29 adds role_judgements.musts and broken; the judge's input_key carries a format version so earlier judgements are made again. The reader lists the must-haves with their verdicts and the cap.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-03 22:40:54 +02:00
WolfandClaude Opus 5.5 64fecf57ad Fold the CV match plan into the decisions and delete it
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-03 22:28:21 +02:00
WolfandClaude Sonnet 5.5 585c332d74 Make the three CV match weights a per-agent setting, applied when a role is served
cv_skills, cv_experience and cv_career live in agents.weights (30, 25, 30 by
default), edited on the Search agents page beside fit and recency. cv_score is
worked out at read time in the feed order, the row and the reader, so changing
them asks the model for nothing and leaves the judge's input_key alone.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-03 22:25:44 +02:00
WolfandClaude Opus 5.5 810b41b733 Add a must-haves and deal-breakers text to the CV match plan
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-03 21:48:47 +02:00
WolfandClaude Opus 5.5 ed6d641373 Plan the CV match: a judge stage and a third sort, a language flag and a deadline filter
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-03 21:43:39 +02:00
WolfandClaude Opus 5.5 8c139f7f5a Fold the search providers plan into the decisions and delete it
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-03 12:32:07 +02:00
WolfandClaude Opus 5.5 9b3b4ea57f Add WP5 to the search providers plan: text for a search reader with detail
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-03 12:23:13 +02:00
WolfandClaude Opus 5.5 018a1ad895 Plan three more search providers: freehire, Jobnet and Akademikernes Jobbank
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-03 12:12:49 +02:00
WolfandClaude Opus 5.5 c497fe965d Keep a description's headings, bullets and paragraphs, and draw them in the reader
The readers flattened a posting's HTML into one line with strip_html, so
sections and bullets ran together. _shared.html_text gives a line per
paragraph, '• ' before a list item and '## ' before a heading; every
reader's description goes through it, sections read per role get their
own headings, and the reader pane draws the lines as h3, ul and p.
Excerpts leave the marks out.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-09-30 21:40:12 +02:00
WolfandClaude Opus 5.5 17e7615599 Order a role's buttons ▼ ● ▲ ★ and like a role when it is saved
The company control follows as ⊘ ▲ ★, so like and favourite sit together.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-09-30 21:06:11 +02:00
WolfandClaude Opus 5.5 bf6ef72b15 Fold the remote area plan into the decisions and delete it
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-09-30 16:29:16 +02:00
WolfandClaude Fable 5.1 a211cf7414 Fold the review of 2026-09-30 into the decisions and open items, and delete the plan
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-09-30 16:11:27 +02:00
WolfandClaude Fable 5.1 dd0eee9ebb Merge main (the R3 bug fixes) into wp-review-docs
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-09-30 15:46:52 +02:00
WolfandClaude Fable 5.1 377cbf7e2d Correct docs/plan-hosts-wave-2.md: seed versus live counts, dead references, WP9 scope and ordering
State that the 76 is the seed's count, name the survey files that do not exist, make WP9 own _runner.py because detail cannot fill a title until DETAIL_FIELDS has it, and make it wait for plan-remote's WP-A and the review's R1, which own the same files.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-09-30 15:44:24 +02:00
WolfandClaude Fable 5.1 1ccfec97ea Correct docs/filter-survey.md against the readers and say what F2 landed as
State which hosts the survey covers and which it leaves out, fix the Workday and Oracle facet names, the All countries versus All roles sentence and the places-in-reach claim, mark the Workable, Teamtailor and Jibe query rows as not built, and reword the plan voice and first person.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-09-30 15:44:06 +02:00
WolfandClaude Fable 5.1 e1b465f405 Refresh open.md: numbers, boilerplate drift, HR Manager, empty slugs, and the review's unbuilt items
Update the stale counts, state which boilerplate copies lag and by what, record HR Manager as attached against its robots record, add Ashby to the empty-slug item, the SmartRecruiters case question and the seed lag, and add a section for what the review of 2026-09-30 does not build.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-09-30 15:43:27 +02:00
WolfandClaude Fable 5.1 597ce9ca39 Mark the decisions that later ones superseded and fix stale words in decisions.md
Each older paragraph that a later decision contradicts (cascade, Jobindex completeness and link-out, host counts, HR Manager and the three Workday companies, summaries, seniority order, trusted dates, tabs, the 175 count) gets one sentence naming what replaced it. The Jobindex robots record names sort=. Em-dashes and a role called a job are reworded.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-09-30 15:42:40 +02:00
WolfandClaude Opus 5.5 a1b13e8bc5 Plan: the plain remote gate stays, remote_from is the narrow line
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-09-30 15:29:44 +02:00