Skip to content

docs: checked against Claude Code 2.1.316, and its resume warning - #52

Merged
rbartoli merged 4 commits into
mainfrom
docs/claude-code-2-1-310
Oct 31, 2026
Merged

rbartoli merged 4 commits into
mainfrom
docs/claude-code-2-1-310

Conversation

@rbartoli

@rbartoli rbartoli commented Oct 24, 2026 •

Copy link
Copy Markdown
Owner

What changes

  • CI tests on Claude Code 2.1.316, and the README's coverage table says "Checked against v2.1.316". Going through the changelog for 2.1.296–2.1.316, I found no change that alters a coverage row.
  • Claude Code's own resume warning (2.1.304) is described in the README. It shows only from a minimum share of the 5-hour limit, so the guard's dormant question, from 150k tokens, stays.
  • Idle compaction now runs in more places (2.1.314): VS Code, Claude Desktop's Code tab, subscribers with telemetry off, and Bedrock, Vertex, Foundry and LLM gateways with the one-hour cache. The README's list of what the dormant question still covers now reads "claude -p and SDK sessions" instead of "other non-terminal sessions".
  • CLAUDE_CODE_SUBAGENT_EFFORT (2.1.316) is named in "Native limits still matter". It makes each agent cheaper but doesn't limit how many agents run.
  • main is merged in after 1.1.2, and the changelog entry sits under Unreleased.

Checked for 2.1.311–2.1.316

  • "Usage limit messages call the 5-hour window the '5-hour limit'" (2.1.312): core/lockout.ts:10 already matches 5-hour limit.
  • No mod API change in the changelog: plugin validate --strict and plugin test (100 pass) both run clean on 2.1.316.
  • Nothing native replaces a guard feature. There's still no cap on agent starts or concurrency across sessions, and no question before heavy prompts.

Test plan

  • claude plugin test . on 2.1.316
  • claude plugin validate . --strict on 2.1.316
  • CI green on the 2.1.316 pin

https://claude.ai/code/session_01GXeouKcCrPifmnnBAmRCF4

@rbartoli rbartoli mentioned this pull request Oct 24, 2026
1 of 3 tasks
@rbartoli rbartoli changed the title docs: checked against Claude Code 2.1.310, and its resume warning docs: checked against Claude Code 2.1.316, and its resume warning Oct 31, 2026
* origin/main:
  fix: the guard never submits a prompt, and Cancel no longer doubles it (#56)

# Conflicts:
#	CHANGELOG.md
@rbartoli
rbartoli merged commit cb84530 into main Oct 31, 2026
4 checks passed
@rbartoli
rbartoli deleted the docs/claude-code-2-1-310 branch October 31, 2026 21:12
@rbartoli rbartoli mentioned this pull request Oct 31, 2026
1 of 3 tasks
rbartoli added a commit that referenced this pull request Oct 31, 2026
* origin/next:
  chore: release 1.1.3 (#57)
  docs: checked against Claude Code 2.1.316, and its resume warning (#52)
  fix: the guard never submits a prompt, and Cancel no longer doubles it (#56)
  docs: date 1.1.2 to the day it was released (#55)
  chore: release 1.1.2 (#53)
rbartoli added a commit that referenced this pull request Oct 31, 2026
…estions

* origin/next:
  docs: say the store's records are keyed by the full session id (#43)
  chore: release 1.1.3 (#57)
  docs: checked against Claude Code 2.1.316, and its resume warning (#52)
  fix: the guard never submits a prompt, and Cancel no longer doubles it (#56)
  docs: date 1.1.2 to the day it was released (#55)
  chore: release 1.1.2 (#53)

# Conflicts:
#	TESTING.md
rbartoli added a commit that referenced this pull request Oct 31, 2026
…cords

* origin/next:
  fix: a question shows an agent's name only when it is a plain one (#44)
  docs: say the store's records are keyed by the full session id (#43)
  chore: release 1.1.3 (#57)
  docs: checked against Claude Code 2.1.316, and its resume warning (#52)
  fix: the guard never submits a prompt, and Cancel no longer doubles it (#56)
  docs: date 1.1.2 to the day it was released (#55)
  chore: release 1.1.2 (#53)
rbartoli added a commit that referenced this pull request Oct 31, 2026
…failures

* origin/next:
  fix: check records read back from the store value by value (#45)
  fix: a question shows an agent's name only when it is a plain one (#44)
  docs: say the store's records are keyed by the full session id (#43)
  chore: release 1.1.3 (#57)
  docs: checked against Claude Code 2.1.316, and its resume warning (#52)
  fix: the guard never submits a prompt, and Cancel no longer doubles it (#56)
  docs: date 1.1.2 to the day it was released (#55)
  chore: release 1.1.2 (#53)

# Conflicts:
#	README.md
#	TESTING.md
rbartoli added a commit that referenced this pull request Oct 31, 2026
…d-relayed-prompts

* origin/next:
  fix: say when the store stops saving (#46)
  fix: check records read back from the store value by value (#45)
  fix: a question shows an agent's name only when it is a plain one (#44)
  docs: say the store's records are keyed by the full session id (#43)
  chore: release 1.1.3 (#57)
  docs: checked against Claude Code 2.1.316, and its resume warning (#52)
  fix: the guard never submits a prompt, and Cancel no longer doubles it (#56)
  docs: date 1.1.2 to the day it was released (#55)
  chore: release 1.1.2 (#53)

# Conflicts:
#	TESTING.md
rbartoli added a commit that referenced this pull request Oct 31, 2026
…near-limit

* origin/next:
  fix: scheduled and relayed prompts go through the prompt gate (#47)
  fix: say when the store stops saving (#46)
  fix: check records read back from the store value by value (#45)
  fix: a question shows an agent's name only when it is a plain one (#44)
  docs: say the store's records are keyed by the full session id (#43)
  chore: release 1.1.3 (#57)
  docs: checked against Claude Code 2.1.316, and its resume warning (#52)
  fix: the guard never submits a prompt, and Cancel no longer doubles it (#56)
  docs: date 1.1.2 to the day it was released (#55)
  chore: release 1.1.2 (#53)

# Conflicts:
#	TESTING.md
rbartoli added a commit that referenced this pull request Oct 31, 2026
* origin/next:
  feat: from 95%, wakeups and scheduled jobs are refused (#48)
  fix: scheduled and relayed prompts go through the prompt gate (#47)
  fix: say when the store stops saving (#46)
  fix: check records read back from the store value by value (#45)
  fix: a question shows an agent's name only when it is a plain one (#44)
  docs: say the store's records are keyed by the full session id (#43)
  chore: release 1.1.3 (#57)
  docs: checked against Claude Code 2.1.316, and its resume warning (#52)
  fix: the guard never submits a prompt, and Cancel no longer doubles it (#56)
  docs: date 1.1.2 to the day it was released (#55)
  chore: release 1.1.2 (#53)

# Conflicts:
#	TESTING.md
rbartoli added a commit that referenced this pull request Oct 31, 2026
…side-caps

* origin/next:
  feat: the burn rule looks at time to empty (#49)
  feat: from 95%, wakeups and scheduled jobs are refused (#48)
  fix: scheduled and relayed prompts go through the prompt gate (#47)
  fix: say when the store stops saving (#46)
  fix: check records read back from the store value by value (#45)
  fix: a question shows an agent's name only when it is a plain one (#44)
  docs: say the store's records are keyed by the full session id (#43)
  chore: release 1.1.3 (#57)
  docs: checked against Claude Code 2.1.316, and its resume warning (#52)
  fix: the guard never submits a prompt, and Cancel no longer doubles it (#56)
  docs: date 1.1.2 to the day it was released (#55)
  chore: release 1.1.2 (#53)

# Conflicts:
#	TESTING.md
rbartoli added a commit that referenced this pull request Oct 31, 2026
* origin/next:
  feat: Haiku agents stay out of the running and starts limits (#50)
  feat: the burn rule looks at time to empty (#49)
  feat: from 95%, wakeups and scheduled jobs are refused (#48)
  fix: scheduled and relayed prompts go through the prompt gate (#47)
  fix: say when the store stops saving (#46)
  fix: check records read back from the store value by value (#45)
  fix: a question shows an agent's name only when it is a plain one (#44)
  docs: say the store's records are keyed by the full session id (#43)
  chore: release 1.1.3 (#57)
  docs: checked against Claude Code 2.1.316, and its resume warning (#52)
  fix: the guard never submits a prompt, and Cancel no longer doubles it (#56)
  docs: date 1.1.2 to the day it was released (#55)
  chore: release 1.1.2 (#53)

# Conflicts:
#	TESTING.md
#	hooks/register.ts
rbartoli added a commit that referenced this pull request Oct 31, 2026
* origin/next:
  feat: /agent-guard simulate shows what a rule does (#51)
  feat: Haiku agents stay out of the running and starts limits (#50)
  feat: the burn rule looks at time to empty (#49)
  feat: from 95%, wakeups and scheduled jobs are refused (#48)
  fix: scheduled and relayed prompts go through the prompt gate (#47)
  fix: say when the store stops saving (#46)
  fix: check records read back from the store value by value (#45)
  fix: a question shows an agent's name only when it is a plain one (#44)
  docs: say the store's records are keyed by the full session id (#43)
  chore: release 1.1.3 (#57)
  docs: checked against Claude Code 2.1.316, and its resume warning (#52)
  fix: the guard never submits a prompt, and Cancel no longer doubles it (#56)
  docs: date 1.1.2 to the day it was released (#55)
  chore: release 1.1.2 (#53)

# Conflicts:
#	TESTING.md
#	core/index.ts
#	hooks/register.ts
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant