sandtable-ai

Issues somebody may already be working on are listed last. List everything by date

• plugin: an eval suite for the skills, built from a drafting job with planted traps

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 55/100

  sandtable-ai/sandtable#302 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• plugin: the marketplace installs the skills from main, so they document features the published CLI doesn't have, under the release's version number

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  type:bug

  Difficulty 4/5 3-5 days Newbie friendliness 38/100

  sandtable-ai/sandtable#301 · 2 comments · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• plugin: the skills and failure-triager never use regrade, so a grading-only fix is confirmed with a paid rerun

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  type:docs

  Difficulty 2/5 1-3 hours Newbie friendliness 76/100

  sandtable-ai/sandtable#300 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• plugin: attribution blames the agent when it correctly refuses a request that the records contradict

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  type:docs

  Difficulty 3/5 1-2 days Newbie friendliness 62/100

  sandtable-ai/sandtable#299 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• plugin: the review checklists miss checks that a run on real support chats needed

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  type:docs

  Difficulty 2/5 1-3 hours Newbie friendliness 78/100

  sandtable-ai/sandtable#298 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• init: --sop and --transcripts report no token usage or estimated cost

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:init type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  sandtable-ai/sandtable#297 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• plugin: before a paid init step, the drafting skill should say how many model calls it makes and roughly what they cost

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  type:docs

  Difficulty 3/5 1-2 days Newbie friendliness 57/100

  sandtable-ai/sandtable#296 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• plugin: the skills don't say how to install the Bedrock client, so an agent works out how to run init on Bedrock from sandtable's compiled code

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  type:docs

  Difficulty 2/5 1-3 hours Newbie friendliness 78/100

  sandtable-ai/sandtable#295 · 1 comment · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• `init --transcripts` drafts a task's scenario rows as `given.rows` instead of shared seed rows

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:init priority:p2 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 56/100

  sandtable-ai/sandtable#286 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• `sandtable compare`: show what changed between two stored runs

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:cli priority:p2 type:feature

  Difficulty 2/5 1-3 hours Newbie friendliness 76/100

  sandtable-ai/sandtable#285 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• Leaderboard and `demo --live`: flag critical failures and treat errored trials as `run` does

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:report priority:p2 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  sandtable-ai/sandtable#284 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• Per-tag pass^k floors (`--floor refusal=100`)

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:grading priority:p2 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  sandtable-ai/sandtable#283 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• Exit 1 for a failed gate and 2 for an error, as grep and diff do

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:cli priority:p1 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  sandtable-ai/sandtable#282 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• Fail CI on regressions since a baseline (`--fail-on-regression`)

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:cli priority:p1 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 45/100

  sandtable-ai/sandtable#281 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• init: --transcripts drafts no reference calls, so no drafted task can be proven solvable

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:init type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  sandtable-ai/sandtable#278 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• lint: a reference can't include a call the API should reject, so refusal tasks can't be proven solvable

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:grading type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 58/100

  sandtable-ai/sandtable#274 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• docs: the quickstart's toy agent forgets the MC number, so its transcripts repeat "What's your MC number?"

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  type:docs

  Difficulty 2/5 1-3 hours Newbie friendliness 82/100

  sandtable-ai/sandtable#270 · Opened Oct 21, 2026

  Maintainers usually reply within 1 day

• freight: booking a made-up load ID is reported as a rep-confirm violation

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:grading type:feature

  Difficulty 4/5 1-2 days Newbie friendliness 48/100

  sandtable-ai/sandtable#163 · Opened Oct 20, 2026

  Maintainers usually reply within 1 day

• Answer-shaped docs pages: how-to questions and a glossary

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  type:docs

  Difficulty 2/5 1-3 hours Newbie friendliness 67/100

  sandtable-ai/sandtable#73 · Opened Oct 19, 2026

  Maintainers usually reply within 1 day

• Rehearse Python agents in-process (LiveKit, Pipecat)

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:adapters priority:p2 type:feature

  Difficulty 5/5 Over a week Newbie friendliness 35/100

  sandtable-ai/sandtable#44 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Voice timing metrics spec: turn latency, barge-in and silence

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:voice priority:p2 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  sandtable-ai/sandtable#43 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Multi-session tasks and an outbox for side channels

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:grading priority:p2 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  sandtable-ai/sandtable#42 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Persona × condition matrix

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:grading priority:p2 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 45/100

  sandtable-ai/sandtable#41 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Business metrics per task next to pass^k

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:report priority:p2 type:feature

  Difficulty 5/5 Over a week Newbie friendliness 42/100

  sandtable-ai/sandtable#40 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Grade call outcomes and dispositions

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:grading priority:p2 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 45/100

  sandtable-ai/sandtable#39 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Frozen simulation clock and relative dates

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:grading priority:p2 type:feature

  Difficulty 5/5 Over a week Newbie friendliness 35/100

  sandtable-ai/sandtable#38 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Multi-round negotiation invariants: `max_count` and `monotonic`

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:grading priority:p1 type:feature

  Difficulty 5/5 Over a week Newbie friendliness 35/100

  sandtable-ai/sandtable#36 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Import Retell and ElevenLabs agent exports

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:init priority:p2 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  sandtable-ai/sandtable#35 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• `sandtable import`: draft a domain from a voice-agent platform config

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:init priority:p1 type:feature

  Difficulty 5/5 Over a week Newbie friendliness 38/100

  sandtable-ai/sandtable#34 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Flag stale tasks when the tools, tables or rules they depend on change

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:grading priority:p2 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 45/100

  sandtable-ai/sandtable#33 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Customer sign-off report: rule → evidence → reliability

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:report priority:p1 type:feature

  Difficulty 5/5 Over a week Newbie friendliness 38/100

  sandtable-ai/sandtable#31 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Coverage: which rules and tools the suite actually exercises

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:report priority:p1 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  sandtable-ai/sandtable#30 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Failure attribution: agent, simulated user, task or infrastructure

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:grading priority:p1 type:feature

  Difficulty 5/5 Over a week Newbie friendliness 35/100

  sandtable-ai/sandtable#29 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Nested params: objects, arrays and nullable fields in tool schemas

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:adapters priority:p1 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 52/100

  sandtable-ai/sandtable#26 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Filter tasks by tag (`--tag refusal`)

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:cli good first issue priority:p2 type:feature

  Difficulty 2/5 1-3 hours Newbie friendliness 65/100

  sandtable-ai/sandtable#21 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Add `$regex` to where clauses

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:grading good first issue priority:p2 type:feature

  Difficulty 2/5 1-3 hours Newbie friendliness 78/100

  sandtable-ai/sandtable#20 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Export domains to tau-bench's format

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  priority:p2 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 55/100

  sandtable-ai/sandtable#18 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Voice: rehearse phone agents over audio

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:voice priority:p2 type:feature

  Difficulty 5/5 Over a week Newbie friendliness 35/100

  sandtable-ai/sandtable#17 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• LLM-judged checks: calibrated criteria scored separately from pass/fail

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:grading priority:p2 type:feature

  Difficulty 5/5 Over a week Newbie friendliness 35/100

  sandtable-ai/sandtable#16 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Fault profiles for tools: timeouts, 429s, duplicate deliveries

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:grading priority:p2 type:feature

  Difficulty 5/5 Over a week Newbie friendliness 35/100

  sandtable-ai/sandtable#15 · 1 comment · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• Black-box mode: serve the simulated world as REST and MCP

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:adapters priority:p1 type:feature

  Difficulty 5/5 Over a week Newbie friendliness 35/100

  sandtable-ai/sandtable#14 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day

• GitHub Action that comments pass^k deltas on pull requests

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area:cli priority:p1 type:feature

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  sandtable-ai/sandtable#13 · Opened Oct 18, 2026

  Maintainers usually reply within 1 day