Autonomous Agent · Live Audit Log
The Curator Agent
DepScope's data is continuously curated by an autonomous agent that runs in production, picks its own actions from a whitelisted catalog, and writes them to the database. Everything below is live: iterations, actions, rows affected, reasoning. Public. CC0.
Iterations
278
since launch
Rows curated
801,526
last 24 hours
Errors
0
all-time
Cycle
m
last: 47m ago
How it works
1. Gap analysis
On each iteration the agent runs read-only SQL to inventory the largest data gaps: missing download counts, stale health scores, duplicate notations, date inversions, deps not yet expanded.
2. Action
The agent picks one action from a whitelisted catalog (idempotent only), with bounded row caps. Every decision is recorded with a rationale.
3. Audit
Every action is idempotent, row-capped, and logged with rationale. Cache invalidation, anti-loop cooldown, and DB constraints are part of the safety envelope.
Action mix · last 24 hours
| Action | Count | Rows affected |
|---|---|---|
| expand_vertical_deps | 114 | 351,688 |
| derive_dl_monthly | 56 | 272,011 |
| run_cron | 26 | 26 |
| noop | 13 | 0 |
| derive_latest_version | 8 | 135,000 |
| derive_dl_weekly | 3 | 14,661 |
| merge_cpan_dual_notation | 2 | 95 |
| prune_self_ref_alts | 1 | 0 |
| fix_inverted_dates | 1 | 23,044 |
| triage_health_findings | 1 | 1 |
Recent decisions (50)
#278triage_health_findings47m ago+1 rows11935ms
Open critical/high probe finding present (1) — directive says PICK whenever > 0 since these are real bugs needing roadmap_item triage.
#277expand_vertical_deps1h ago+19 rows13639ms
Vertical deps pending backlog is massive (1,085,676) — extracting data_json.dependencies into the package_dependencies table at 50k/iter delivers the largest single-shot value and unblocks downstream graph queries.
#276expand_vertical_deps1h ago+13 rows13450ms
Vertical deps pending backlog is massive (1.08M) — extracting deps from data_json into the package_dependencies table in one capped batch yields the highest single-iteration value.
#275expand_vertical_deps1h ago+5 rows12563ms
Vertical deps backlog is >1M packages — by far the largest fixable gap, and one capped batch (50k) makes meaningful dent on dependency graph coverage.
#274expand_vertical_deps1h ago14141ms
Vertical deps backlog is over 1M packages — the largest single gap in the DB, and one batch of 50k rows in this iteration delivers the most rows-fixed-per-action of any available choice.
#273expand_vertical_deps1h ago14220ms
Vertical deps backlog is over 1M packages — extracting data_json.dependencies into the package_dependencies table is the highest-leverage single action available, fixing up to 50k rows in one shot.
#272expand_vertical_deps1h ago13083ms
Vertical deps backlog is over 1M pending rows — by far the largest fixable gap, and one iteration moves up to 50k rows from data_json into the package_dependencies table.
#271expand_vertical_deps1h ago12726ms
Vertical deps backlog of 1.08M pending rows is by far the largest fixable gap; one capped iteration extracts up to 50k dependency rows in a single shot.
#270expand_vertical_deps1h ago12579ms
Vertical deps backlog is enormous (>1M pkg pending) and a single 50k-row batch makes large measurable progress on the biggest structural gap.
#269expand_vertical_deps1h ago12783ms
Vertical deps pending backlog exceeds 1M packages — extracting data_json.dependencies into the table closes the largest single structural gap and 50K/iter makes meaningful progress.
#268expand_vertical_deps1h ago13839ms
Vertical deps backlog exceeds 1M packages — single biggest fixable gap, SQL-only and capped at 50k/iter for safe progress.
#267expand_vertical_deps1h ago+3 rows12952ms
Vertical deps backlog is over 1M packages — extracting data_json.dependencies into the relational table delivers the highest one-shot value this iteration.
#266expand_vertical_deps1h ago12629ms
Vertical deps backlog of 1.08M is by far the largest gap and can extract 50K rows in one shot from existing data_json with no network cost.
#265expand_vertical_deps1h ago12350ms
Vertical deps backlog is over 1 million packages — by far the largest single fixable gap, and this action processes 50k/iter for maximum throughput per run.
#264expand_vertical_deps1h ago12444ms
Vertical deps backlog is over 1M packages — by far the largest actionable gap, and one cap-50k iteration meaningfully chips away at it.
#263expand_vertical_deps1h ago12554ms
Vertical deps pending backlog is 1.08M packages — by far the largest gap; one iteration capped at 50k rows makes meaningful progress on the most under-utilized data.
#262expand_vertical_deps1h ago13369ms
Vertical deps backlog is >1M pending — extracting data_json.dependencies into the package_dependencies table at the 50k/iter cap delivers the largest single-shot coverage gain available.
#261expand_vertical_deps1h ago+10 rows15745ms
Vertical deps backlog is massive (1.085M pending) — extracting data_json.dependencies into the package_dependencies table at 50k/iter cap delivers the highest single-shot value with no destructive risk.
#260expand_vertical_deps1h ago13015ms
Vertical deps pending backlog is massive (1,085,622 rows) — extracting data_json.dependencies into package_dependencies table at 50k/iter cap delivers the largest single-shot value and unlocks dependency graph queries.
#259expand_vertical_deps1h ago13349ms
Vertical deps backlog is >1M packages — extracting data_json.dependencies into the package_dependencies table at 50k/iter delivers the largest one-shot value this iteration.
#258expand_vertical_deps1h ago+7 rows13832ms
Vertical deps backlog >1M pending — single SQL action processes 50k rows per iteration for maximum throughput on the largest gap.
#257expand_vertical_deps1h ago13515ms
Vertical deps backlog is massive at 1.08M pending — extracting data_json.dependencies into the package_dependencies table delivers the highest single-iteration value.
#256expand_vertical_deps1h ago+17 rows13524ms
Vertical deps backlog is over 1M pending rows — by far the largest fixable gap, and one batch (cap 50k/iter) makes meaningful progress on the dependency graph in a single shot.
#255expand_vertical_deps1h ago14231ms
Vertical deps backlog is massive (1.08M pending) — extracting data_json.dependencies into the package_dependencies table is the highest-leverage one-shot action this iteration (cap 50K/iter).
#254expand_vertical_deps1h ago+3 rows13801ms
Vertical deps backlog of 1,085,599 pending rows is by far the largest fixable gap; one batch capped at 50K rows materially grows the package_dependencies table in a single shot.
#253derive_latest_version1h ago+5,000 rows24365ms
141,571 packages have data_json.latest_version available but the column is NULL — a single SQL UPDATE closes the largest one-shot column-level gap in the DB.
#252expand_vertical_deps1h ago13559ms
Vertical deps backlog is over 1M pending rows — extracting data_json.dependencies into the package_dependencies table in 50k-row batches yields the largest one-shot value this iteration.
#251expand_vertical_deps1h ago13995ms
Vertical deps pending backlog is >1M packages — by far the largest single fixable gap, and one iteration capped at 50k rows makes meaningful progress on dependency graph coverage.
#250expand_vertical_deps1h ago13592ms
Vertical deps backlog is massive (1.08M pending) — extracting data_json.dependencies into the package_dependencies table closes the largest single gap in one cap-bound batch.
#249expand_vertical_deps1h ago+3 rows14275ms
Vertical deps pending backlog is over 1M rows — extracting data_json.dependencies into the table yields the largest single-iteration coverage gain.
#248expand_vertical_deps1h ago13990ms
Vertical deps backlog is over 1M packages — extracting dependencies into the relational table delivers the largest one-shot value among all gaps.
#247expand_vertical_deps1h ago+10 rows14100ms
Vertical deps pending backlog is >1M (1,085,551), by far the largest fixable gap — one iteration extracts up to 50K dependencies in a single shot, far more value than any other action.
#246expand_vertical_deps1h ago13212ms
Vertical deps pending backlog is massive (1,085,542) — extracting data_json.dependencies into the package_dependencies table in one 50k-cap pass yields the highest single-iteration value.
#245expand_vertical_deps1h ago13971ms
Vertical deps backlog is over 1M rows, vastly exceeding any other gap — extracting 50k deps/iter delivers the largest single-action value.
#244expand_vertical_deps1h ago+4 rows15440ms
Vertical deps backlog is over 1M packages — extracting data_json.dependencies into the table at 50k/iter cap is the highest-leverage single action.
#243expand_vertical_deps1h ago+4 rows12883ms
Vertical deps backlog is >1M packages — running this extraction yields the largest single-iteration value by populating package_dependencies for up to 50k packages.
#242expand_vertical_deps1h ago+11 rows16018ms
Vertical deps backlog is over 1M packages — extracting data_json.dependencies into the table at 50k/iter cap delivers the largest single-shot value of any available action.
#241expand_vertical_deps2h ago13016ms
Vertical deps backlog is over 1M packages — extracting data_json.dependencies into the package_dependencies table at 50k/iter delivers the largest single-shot value this cycle.
#240expand_vertical_deps2h ago13872ms
Vertical deps backlog at 1.08M is by far the largest fixable gap — one iteration extracts up to 50k dependency rows, dwarfing all other actions in value per run.
#239expand_vertical_deps2h ago+5 rows16405ms
Vertical deps backlog is over 1M packages — by far the largest fixable gap, and one SQL pass extracts up to 50k pkg deps per iteration with no network cost.
#238expand_vertical_deps2h ago+2 rows14824ms
Vertical deps backlog is over 1M packages — by far the largest gap; one capped iteration extracts up to 50k rows of dependency graph data in a single shot.
#237expand_vertical_deps2h ago+1 rows14437ms
Vertical deps backlog at 1.085M is by far the largest fixable gap and a single iteration processes up to 50K packages, delivering massive coverage gain in one shot.
#236expand_vertical_deps2h ago13026ms
Massive 1.08M backlog of pending vertical dependency extractions — single SQL action processes up to 50000 pkg/iter, by far the highest-leverage gap-closer this iteration.
#235expand_vertical_deps2h ago13876ms
Vertical deps backlog is massive (>1M pending) — extracting data_json.dependencies into the package_dependencies table delivers the highest single-iteration value.
#234expand_vertical_deps2h ago+4 rows14267ms
Vertical deps backlog is over 1M pending — extracting dependencies into the table provides the highest single-iteration coverage gain.
#233expand_vertical_deps2h ago+53 rows13564ms
Vertical deps backlog is massive (1.08M pending) — extracting data_json.dependencies into the table fixes the most rows in one shot.
#232expand_vertical_deps2h ago+20 rows14450ms
Vertical deps backlog is massive at 1.08M pending — extracting data_json.dependencies into the table fills a huge gap in one capped batch (50k pkg/iter).
#231expand_vertical_deps2h ago+17 rows14215ms
Vertical deps backlog is massive (1.08M pending) and a single SQL extraction can process up to 50k packages per iteration, delivering the largest one-shot value.
#230expand_vertical_deps2h ago+3 rows13405ms
Vertical deps backlog is over 1M packages — extracting data_json.dependencies into package_dependencies table at 50k/iter cap delivers the largest one-shot value this iteration.
#229expand_vertical_deps2h ago13535ms
Vertical deps backlog is massive (>1M pending) — extracting data_json.dependencies into the package_dependencies table delivers the highest one-shot value of any available action.