Skip to content

temporal — Issues Report

Generated: 2026-07-07 CLI version: temporal version 1.7.2 (Server 1.31.1, UI 2.49.1) Scope: critical Findings in scope: 22 failure modes

Observed Bugs (from evaluation notes)

These were witnessed directly when running checks against this CLI.

§2 candidate — --output json does not apply to validation failures

Discovered during: §2 evaluation — 2026-07-07 Symptom: temporal --output json workflow start without required flags prints usage text to stdout and prose errors to stderr. Impact: Agents expecting JSON must switch parsers on failure and can accidentally ingest usage text as data. Trigger: /opt/homebrew/bin/temporal --output json workflow start

§11 candidate — Timeout-like network failures collapse to exit 1 and prose

Discovered during: §11 evaluation — 2026-07-07 Symptom: An unroutable server with 2s command/client timeouts exits 1 with Error: program interrupted. Impact: Agents cannot distinguish timeout, cancellation, and other failures from the exit code or JSON body. Trigger: /opt/homebrew/bin/temporal --output json --address 203.0.113.1:7233 --client-connect-timeout 2s --command-timeout 2s workflow list

§23 candidate — Destructive commands lack dry-run/effect contracts

Discovered during: §23 evaluation — 2026-07-07 Symptom: workflow delete --dry-run is rejected as an unknown flag, and query deletes require --yes but return only a batch job ID. Impact: An agent cannot get a machine-readable would-delete scope or distinguish created/noop/effect outcomes. Trigger: /opt/homebrew/bin/temporal workflow delete --dry-run --workflow-id missing-audit-id

§74 candidate — No machine-readable credential scope declaration

Discovered during: §74 evaluation — 2026-07-07 Symptom: --schema, manifest, and check-permissions probes all fail as unknown. Impact: Agents cannot select minimally scoped credentials before invocation. Trigger: /opt/homebrew/bin/temporal check-permissions --for workflow list

Failure-Mode Gaps (score 0–2, sorted: score asc, severity desc; ?/3 entries listed last)

These are not confirmed bugs but verified gaps. ?/3 entries are included at the end with behavior unknown.

§1 — Exit Codes & Status Signaling [Critical · score 0/3]

What fails: Missing args, not-found, and network/timeout failures all exited 1; no declared semantic code table or JSON error exit_code field was observed. Frequency: Very Common Token/time cost when it triggers: Token Spend: High · Time: High Workaround exists: Yes

§13 — Partial Failure & Atomicity [Critical · score 0/3]

What fails: A deliberate bad batch/query path returned a single prose error with no partial, completed_steps, failed_step, rollback, or resume token. Frequency: Common Token/time cost when it triggers: Token Spend: High · Time: High Workaround exists: Yes

§25 — Prompt Injection via Output [Critical · score 0/3]

What fails: User-controlled/local config data is returned as raw JSON values with no trust boundary, envelope, or trusted: false metadata. Frequency: Situational Token/time cost when it triggers: Token Spend: High · Time: High Workaround exists: Yes

§74 — Credential Scope Declaration Absence [Critical · score 0/3]

What fails: --schema, manifest, and check-permissions --for ... are absent; command entries expose no required_scopes field. Frequency: Common Token/time cost when it triggers: Token Spend: Low · Time: Medium Workaround exists: Yes

§2 — Output Format & Parseability [Critical · score 1/3]

What fails: --output json exists and success paths return JSON, but validation and connection failures print prose/usage text instead of an ok/data/error envelope. Frequency: Very Common Token/time cost when it triggers: Token Spend: High · Time: Medium Workaround exists: Yes

§11 — Timeouts & Hanging Processes [Critical · score 1/3]

What fails: --command-timeout and --client-connect-timeout work, but timeout/network failures exit 1 with Error: program interrupted rather than structured TIMEOUT JSON. Frequency: Common Token/time cost when it triggers: Token Spend: High · Time: Critical Workaround exists: Yes

§12 — Idempotency & Safe Retries [Critical · score 1/3]

What fails: Workspace-local env set was repeatable, and workflow start has ID conflict/reuse policies, but no --idempotency-key or response effect contract exists. Frequency: Common Token/time cost when it triggers: Token Spend: High · Time: High Workaround exists: Yes

§23 — Side Effects & Destructive Operations [Critical · score 1/3]

What fails: Query-based destructive commands have --yes prompt bypass and JSON mode refuses prompts, but destructive commands expose no --dry-run, danger_level, or effect field. Frequency: Common Token/time cost when it triggers: Token Spend: Medium · Time: High Workaround exists: Yes

§24 — Authentication & Secret Handling [Critical · score 1/3]

What fails: Temporal accepts API keys through --api-key, which exposes secrets in process arguments; a debug probe did not echo the test key in captured output. Frequency: Common Token/time cost when it triggers: Token Spend: Medium · Time: Medium Workaround exists: Yes

§34 — Shell Injection via Agent-Constructed Commands [Critical · score 1/3]

What fails: Some enum-style flags reject invalid values, but a %2F-encoded environment name was accepted and errors are unstructured; no agent-hardening declaration exists. Frequency: Common Token/time cost when it triggers: Token Spend: High · Time: High Workaround exists: Yes

§42 — Debug / Trace Mode Secret Leakage [Critical · score 1/3]

What fails: Debug logging did not echo the test API key, but secrets can still be supplied on argv and no schema marks sensitive fields. Frequency: Situational Token/time cost when it triggers: Token Spend: Low · Time: Low Workaround exists: Yes

§43 — Tool Output Result Size Unboundedness [Critical · score 1/3]

What fails: List commands expose --limit/--page-size, but outputs lack meta.truncated, total byte counts, or a global --max-output guard. Frequency: Common Token/time cost when it triggers: Token Spend: Critical · Time: High Workaround exists: Yes

§50 — Stdin Consumption Deadlock [Critical · score 1/3]

What fails: Commands with missing required input failed quickly with usage/prose and did not block, but no structured STDIN_REQUIRED code or hint was emitted. Frequency: Common Token/time cost when it triggers: Token Spend: High · Time: Critical Workaround exists: Yes

§60 — OS Output Buffer Deadlock [Critical · score 1/3]

What fails: server start-dev emitted startup lines, but no JSON heartbeat or progress protocol for long-running commands was observed. Frequency: Common Token/time cost when it triggers: Token Spend: High · Time: Critical Workaround exists: Yes

§61 — Bidirectional Pipe Payload Deadlock [Critical · score 1/3]

What fails: Workflow input commands provide --input-file, but stdin size limits and STDIN_TOO_LARGE overflow errors were not observed. Frequency: Situational Token/time cost when it triggers: Token Spend: High · Time: Critical Workaround exists: Yes

§71 — Non-Interactive Installation Absence [Critical · score 1/3]

What fails: The binary is installed and --version works, but this audit workspace has no AGENTS.md/README documenting a non-interactive, idempotent install command. Frequency: Common Token/time cost when it triggers: Token Spend: Low · Time: Critical Workaround exists: Yes

§10 — Interactivity & TTY Requirements [Critical · score 2/3]

What fails: Local mutating config commands complete with stdin closed, and JSON destructive query commands require prompt bypass; no universal schema declares interactive paths. Frequency: Common Token/time cost when it triggers: Token Spend: High · Time: Critical Workaround exists: Yes

§45 — Headless Authentication / OAuth Browser Flow Blocking [Critical · score 2/3]

What fails: Temporal uses API-key/config flags and no browser OAuth flow was observed; auth failures are not exposed as structured AUTH_REQUIRED/auth_methods envelopes. Frequency: Common Token/time cost when it triggers: Token Spend: High · Time: Critical Workaround exists: Yes

§64 — Headless Display and GUI Launch Blocking [Critical · score 2/3]

What fails: server start-dev documents --headless and prints service/UI URLs instead of opening a browser; no schema declares GUI/headless behavior. Frequency: Common Token/time cost when it triggers: Token Spend: High · Time: Critical Workaround exists: Yes

§53 — Credential Expiry Mid-Session [Critical · score ?/3]

What fails: Could not safely create or mock an expired Temporal credential in this environment; behavior remains unverified. Frequency: Common Token/time cost when it triggers: Token Spend: High · Time: High Workaround exists: Partial

Passing (score 3/3 — safe to use without special handling)

  • §37 — REPL / Interactive Mode Accidental Triggering
  • §62 — $EDITOR and $VISUAL Trap

Risk Summary

Category Count §N list
Observed bugs 4 §2, §11, §23, §74
Score 0 — complete failure 4 §1, §13, §25, §74
Score 1 — major gap 12 §2, §11, §12, §23, §24, §34, §42, §43, §50, §60, §61, §71
Score 2 — minor gap 3 §10, §45, §64
Score 3 — passing 2 §37, §62
Indeterminate (?/3) 1 §53

Highest-risk combination: Temporal has useful JSON success output and safe prompt behavior in places, but failures collapse to prose/exit 1 while destructive, auth, and scope contracts remain under-specified for autonomous agents.