Findings from spikes, investigations, and extractions — risks and unknowns retired before they are built on.
An analysis document (AN-xxx) records what was investigated, what was found, and what it means for plans and changes. Findings are historical records: written once at the end of a spike, then never rewritten — a later spike that learns more writes a new document. Plans and proposals cite findings instead of re-arguing them.
An incident where the corpus was green but later evidence contradicted it carries reality: contradicted and links every capability or acceptance criterion whose claim failed, in addition to the carriers that failed to prevent the incident. clue validate --reality-gaps derives the affected-capability view from those edges; it is not a production telemetry channel.
- AN-004 — Cliewen extraction report ·
active— What the repository's prose became, what was minted, and what the dispositions say about the state of its evidence. - AN-001 — Single-battle score deltas are uninformative for low-scoring bots ·
active— Why the current flagged bots are a reason to look rather than a baseline to measure against. - AN-002 — Tank Royale exposes no deterministic random seed ·
active— Classic can make a battle reproducible and Tank Royale cannot, which rules out the strongest class of conformance test. - AN-003 — The harness is Python, which is not a supported evidence carrier ·
active— Why the harness's working behaviour stays draft rather than being badged Human, and the three routes out. - AN-005 — Gun and radar turn-remaining were swapped in the robot status ·
active— A silent defect in a positional constructor call, found by the first unit test this repository ever ran. - AN-006 — A robot's own death still does not reach onDeath under the bridge ·
active— WhetherEVT-004holds: a robot's own death reaches its death handler. - AN-004-extraction/
- AN-007 — Remaining turn had opposite signs depending on which path a robot read it through ·
active— Two conversion paths for one quantity disagreed, and the test written alongside AN-005 had certified the wrong one. - AN-008 — Tank Royale declares more round winners than the battle has rounds ·
active— Round outcomes, noticed while measuringAN-006. - AN-009 — The Tank Royale server never sends a death event to any bot, which is why onDeath never fires ·
active— The cause ofAN-006, which measured thatonDeathis never called under the bridge and named a suspect it could not confirm: the Bot API event queue's age and criticality filter. - AN-010 — The unit tier's evidence tags mostly hold — two small gaps, and why the failure mode CH-003 found is concentrated elsewhere ·
active—G-002's extension from the conformance tier, which it audited, to the unit tier: every test underrobocode-api/src/test/java/dev/robocode/tankroyale/bridge/against theCAP-003… - AN-011 — M-002 leaves one lower-score bot within the band and names causes for the other two ·
active— The three lower-score candidates were re-measured at officialroborumbleparameters: AetosFirstBot is within the band, while CodaFirst and Bl4ck remain lower with named lifecycle and startup/physics causes. - AN-012 — M-003 finds one bridge-only lifecycle failure and no reproducible higher-score gap in the other two candidates ·
active— The three higher-score candidates were re-measured at officialroborumbleparameters: ScalarR has a named bridge-only lifecycle failure, while Ar1 and BasicSurfer are within the band. - AN-013 — Tank Royale's native team model corresponds closely enough to classic's to map onto directly ·
active—docs/capabilities/CAP-006-team-robot-support/design.mdleft one decision open before anyM-005implementation: does Tank Royale's team concept correspond closely enough to classic Robocode's to… - AN-015 — The first melee registry cases are contaminated by fixed-opponent failures ·
active— Why the first official melee observations cannot yet be read as independent subject failures, and which bridge gaps remain separate investigations.