Skip to content
Proposals/Register workspace-scoped scheduled tasks and du
proposaldeveloperP5Worth a lookCline

Register workspace-scoped scheduled tasks and durable todo agendas in Cline

Cline SDK v0.0.76 lets agents create scheduled tasks and a durable todo agenda scoped to the workspace, generate images mid-turn with history persistence, and load skill slash commands via the skills tool so transcripts stay clean.

Why this loop

Cline SDK v0.0.76 lets agents create and manage scheduled tasks plus a durable todo agenda, with schedules scoped to the workspace that registered them (Windows agenda spec watcher no longer crashes when the specs dir is an 8.3 short path). Image-capable models can generate images mid-turn; those images persist in session history and exports, and image/voice models are removed from chat model pickers. Skill slash commands load through the skills tool instead of pasting SKILL.md into the user message, so the transcript stores /my-skill ..., instructions arrive once, and workflows still expand textually. Same cut restores dropped provider-executed tool activity as observational events, re-delivers PreToolUse contextModification as a hidden <hook_context> block, awaits PostToolUse hooks (120s) and honors contextModification/cancel, routes spaced { command } with no args through the shell, and fail-fasts PowerShell.

Proposed actions

  1. Bump Cline SDK to sdk/sdk/v0.0.76 and have the agent create scheduled tasks plus a durable todo agenda registered against the current workspace only—do not use a process-global schedule store.
  2. On an image-capable model, generate an image during a turn and assert the artifact is present in session history and in the session export; keep image and voice models out of the chat model picker.
  3. Stop inlining SKILL.md into the user message for slash commands. Invoke the skills tool so the persisted transcript records /my-skill ..., skill instructions arrive once, and workflow textual expansion is left unchanged.
  4. Wire hooks to the new contract: re-deliver PreToolUse contextModification on the next engine turn as a <hook_context> block stamped with tool name and call id (hidden from user transcripts); await PostToolUse hooks for 120s and apply their contextModification and cancel controls.
  5. For run_commands, when the payload is structured { command } with a space and no args key, route through the shell to avoid ENOENT; run PowerShell with $ErrorActionPreference='Stop' so a per-item pipeline error fails the command instead of flooding stderr and reporting success.

Agent prompt

Paste into your agent or query via MCP (get_agent_prompt) — free, no extra AI cost

Cline / .clinerules task

DevAgentRadar → Cline

You are helping me adopt a real coding-assistant change. Work only from the facts below. Do not invent features.

Context

Assistant: Cline Proposal: Register workspace-scoped scheduled tasks and durable todo agendas in Cline Summary: Cline SDK v0.0.76 lets agents create scheduled tasks and a durable todo agenda scoped to the workspace, generate images mid-turn with history persistence, and load skill slash commands via the skills tool so transcripts stay clean. Primary source: https://github.com/cline/cline/releases/tag/sdk/sdk/v0.0.76

Why it matters

Cline SDK v0.0.76 lets agents create and manage scheduled tasks plus a durable todo agenda, with schedules scoped to the workspace that registered them (Windows agenda spec watcher no longer crashes when the specs dir is an 8.3 short path). Image-capable models can generate images mid-turn; those images persist in session history and exports, and image/voice models are removed from chat model pickers. Skill slash commands load through the skills tool instead of pasting SKILL.md into the user message, so the transcript stores /my-skill ..., instructions arrive once, and workflows still expand textually. Same cut restores dropped provider-executed tool activity as observational events, re-delivers PreToolUse contextModification as a hidden <hook_context> block, awaits PostToolUse hooks (120s) and honors contextModification/cancel, routes spaced { command } with no args through the shell, and fail-fasts PowerShell.

Suggested actions

  1. Bump Cline SDK to sdk/sdk/v0.0.76 and have the agent create scheduled tasks plus a durable todo agenda registered against the current workspace only—do not use a process-global schedule store.
  2. On an image-capable model, generate an image during a turn and assert the artifact is present in session history and in the session export; keep image and voice models out of the chat model picker.
  3. Stop inlining SKILL.md into the user message for slash commands. Invoke the skills tool so the persisted transcript records /my-skill ..., skill instructions arrive once, and workflow textual expansion is left unchanged.
  4. Wire hooks to the new contract: re-deliver PreToolUse contextModification on the next engine turn as a <hook_context> block stamped with tool name and call id (hidden from user transcripts); await PostToolUse hooks for 120s and apply their contextModification and cancel controls.
  5. For run_commands, when the payload is structured { command } with a space and no args key, route through the shell to avoid ENOENT; run PowerShell with $ErrorActionPreference='Stop' so a per-item pipeline error fails the command instead of flooding stderr and reporting success.

Config surfaces this release may change

  • skills — check your repo before applying
  • hooks — check your repo before applying

After you finish

Do not report this as applied to DevAgentRadar. You cannot write the visitor's loop.

Tell the human: open https://devagentradar.com/proposals/cline-sdk-sdk-v0-0-76-register-workspace-scoped-scheduled-tasks-and-dura and mark Applied, Skipped, or Failed. Proposal id: c8523459-fa2c-4e93-b22f-441db6419ad3

Your job

  1. Restate the change in one sentence.
  2. Propose a minimal plan for my repo (or a throwaway pilot).
  3. Implement only what I approve; prefer small diffs and tests.
  4. Call out risks (permissions, breaking APIs, cost).

Start by confirming you understood the proposal.

agentmodelclibreakingRelease source ↗

Your loop

This browser · no sign-in · not shared as “you”

After you run the prompt

Only you can mark this. Agents cannot write your loop.

Your decision stays on this device. A public tally appears after a few votes.

Originating release signal

ClineSDK 0.0.76Aug 21, 2026

SDK v0.0.76

Added model-driven image generation. · Models that support it can generate images during a turn, and generated images are persisted in session history and exports · Agents can now create and manage scheduled tasks and a durable todo agenda. · Schedules are scoped to the workspace that registered them · Skill slash commands now load through the skills tool instead of being pasted into the user message. · The persisted transcript records what you typed (/my-skill ...) rather than the whole SKILL.md body, and skill instructions arrive once instead of twice. · Workflows keep textual expansion · +17 more changes
Verified excerpt — the source's own words
  • Added model-driven image generation. Models that support it can generate images during a turn, and generated images are persisted in session history and exports
  • Agents can now create and manage scheduled tasks and a durable todo agenda. Schedules are scoped to the workspace that registered them
  • Skill slash commands now load through the skills tool instead of being pasted into the user message. The persisted transcript records what you typed (/my-skill ...) rather than the whole SKILL.md body, and skill instructions arrive once instead of twice. Workflows keep textual expansion
  • Fixed provider-executed tool activity being dropped entirely — every tool the Claude Code provider ran inside its own session modified the workspace with no tool activity in runtime events, transcripts, or the UI. These now surface as observational tool events
  • Fixed PreToolUse hook contextModification never reaching the model on the next engine; it is delivered again as a <hook_context> block stamped with the tool name and call id, hidden from user-facing transcripts

Excerpt ends here — this release continues at the source ↗.

Primary source ↗