didbot
Identities for entities did.bot
atproto agent llm did agentic did.bot
Star 15
Fork 0
Watch
Open Closed
New
blobs: the getBlob stall timer restarts on every chunk, so four slow readers still hold all 32 blob-read permits for over half an hour #243
Closed permadeath.com
· · 1 comment
serve: listRecords has no address bound and reads a page of up to 100 MB inline, so a few callers can stall or exhaust the server #242
Closed permadeath.com
· · 1 comment
oauth: one app can fill its account's DPoP share, refusing the account's other apps and spending their sign-in codes #241
Closed permadeath.com
· · 1 comment
agentd: didbot oauth sends its token to any socket at the path, and prints the reply raw into the agent's context #240
Closed permadeath.com
· · 1 comment
oauth: an app's made-up scope atoms reach the agent's model verbatim, beside the approve command #239
Closed permadeath.com
· · 1 comment
oauth: an authorization code issued before an e-stop Revoke still mints a fresh login after the release #237
Closed permadeath.com
· · 1 comment
pds: the commit size gate misses the sync spec's limits: a record over 1,000,000 bytes is accepted, and tree nodes are not counted #236
Closed permadeath.com
· · 1 comment
pds: one applyWrites can make 3,200 outbound DID lookups, because the per-write bound applies to each write in the batch #235
Closed permadeath.com
· · 2 comments
records: one caller-chosen record key that looks like a TID moves every key the server mints, for every account #234
Closed permadeath.com
· · 1 comment
pds: a record that names many DIDs costs seconds of CPU before any limit, and stalls every route #232
Closed permadeath.com
· · 2 comments
pds: an applyWrites refused partway keeps its earlier ops, after reporting that nothing was written #233
Closed permadeath.com
· · 1 comment
tls: every boot needs Let's Encrypt to answer, even with a valid certificate on disk #231
Closed permadeath.com
· · 1 comment
blobs: 32 slow readers block getBlob for everyone, and queued requests then stall the whole server #229
Closed permadeath.com
· · 2 comments
serve: an open firehose stream holds every restart to the full drain #228
Closed permadeath.com
· · 3 comments
test: write_scale's seeded records never reach the built repository #223
Closed permadeath.com
· · 1 comment
pds: the first commit after a restart sends the whole repository on the firehose #222
Closed permadeath.com
· · 2 comments
serve: a stalled firehose consumer keeps its slot and pins the ring #220
Closed permadeath.com
· · 2 comments
serve: the sync-read budget makes a full indexer pass take 13 min at 10x #219
Closed permadeath.com
· · 2 comments
pds: memory grows with every write between restarts #218
Closed permadeath.com
· · 3 comments
pds: every OAuth refresh rewrites and fsyncs the whole grants file #217
Closed permadeath.com
· · 2 comments
ci: the runner never fetches the MST corpus its tests need #214
Closed permadeath.com
· · 1 comment
serve: one account can fill the DPoP replay window and lock every other account out #213
Closed permadeath.com
· · 2 comments
deploy-pds: every release replaces the ECR lifecycle policy, so a routine deploy needs --allow-replace #212
Closed permadeath.com
· · 1 comment
serve: the stop drain waits out every firehose stream, so a deploy refuses connections for drain_secs #211
Closed permadeath.com
· · 2 comments
serve: send Strict-Transport-Security from the server when its certificate is publicly trusted #210
Closed permadeath.com
· · 1 comment
policy-site: one-pool-per-kind refuses every create on a deployment without those four pools #209
Closed permadeath.com
· · 1 comment
policy: the README's entity-scoped opt-out is not what declared-ai-preference reads #208
Closed permadeath.com
· · 1 comment
ops: clear the empty test accounts on pds.did.bot #207
Closed permadeath.com
· · 1 comment
policy: one account's refused writes spend the floor's server-wide write budget for everyone #204
Closed permadeath.com
· · 2 comments
pds: two durability follow-ups to the layout-stamp fix #201
Closed permadeath.com
· · 4 comments