Skip to content
Agent Orchestration
Skill

/cline-pin-timestamp

Cline's system prompt includes a timestamp that may be recomputed per request, invalidating the system-prompt cache.

BOOST
From plugin
prompt-cache-skills
11613 skills
Install
$ npx -y skills add OnlyTerp/prompt-cache-skills --skill cline-pin-timestamp --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/cline-pin-timestamp

Context preview

The summary Claude sees to decide when to auto-load this skill.

Cline's system prompt includes a timestamp that may be recomputed per request, invalidating the system-prompt cache.

SKILL.md

cline-pin-timestamp.SKILL.md
name: cline-pin-timestamp
description: Cline's system prompt includes a timestamp that may be recomputed per request, invalidating the system-prompt cache.
target_harness: Cline
target_repo: cline/cline
target_files:
  - src/core/prompts/system.ts
target_commit: 65e9727c (verify current at apply time)
estimated_savings: Restores system-prompt cache reads (large; system prompt is typically 5-15k tokens)

Cline: pin the system-prompt timestamp

Target

`src/core/prompts/system.ts` in `cline/cline`. PR #1168 added a current-time field to the system prompt.

Symptom

The system prompt includes the current time. If the time is computed per-request (e.g. `new Date().toISOString()` evaluated at every call site), then the system prompt bytes differ on every request, which invalidates the system-prompt cache breakpoint. The cache write premium gets paid every turn for zero reads, on what's usually the largest single block in the prefix.

PR #1168 was merged despite a reviewer flagging exactly this: > "Just be aware that if the time is being calculated every time the > user sends a request, it invalidates Anthropic's cache."

Verify on the wire: if the system text changes between consecutive calls in the same task, this bug is present.

Fix

Compute the timestamp ONCE at task start, store it on the task, pass through to every system prompt build:

--- a/src/core/prompts/system.ts
+++ b/src/core/prompts/system.ts
@@
-export const SYSTEM_PROMPT = (...args) => `
-  ...
-  Current time: ${new Date().toISOString()}
-  ...
-`
+export const SYSTEM_PROMPT = (args: SystemPromptArgs) => `
+  ...
+  Current time: ${args.taskStartedAt}
+  ...
+`

Then in the task initialization code:

+  this.taskStartedAt = new Date().toISOString()

And pass `this.taskStartedAt` everywhere `SYSTEM_PROMPT` is built.

If date awareness is critical for the agent and you want it to drift during a long session, round to the nearest day or hour rather than the second — that bounds the cache invalidation rate.

Verify

1. Start mitmproxy. 2. Run a Cline task. Wait 10 seconds. Send a second prompt in the same task. 3. Diff the system prompt bytes between the two requests.

  • **Before fix**: bytes differ (timestamp incremented)
  • **After fix**: bytes identical

4. Confirm response `usage.cache_read_input_tokens` covers the system prompt on turn 2.

Background

Anthropic caches the entire prefix UP TO AND INCLUDING the breakpoint. A single byte change before or at the breakpoint invalidates everything. Timestamps are the most common cache killer because they look innocent and few people think to check.

OpenAI's automatic prefix caching has the same vulnerability — same fix applies if Cline targets OpenAI as well.

See [docs/gotchas.md](../../docs/gotchas.md) #5 (byte-identity of tool definitions, same principle for system prompts) and #7 (OpenAI byte-identical prefix requirement).

Full audit: [audits/cline.md](../../audits/cline.md).

Read more
Ships withprompt-cache-skills

Drop-in prompt-caching fixes for the LLM agent harness you use. Point your AI coding agent at this repo and it ships the patches.

Get the whole plugin

Other skills on prompt-cache-skills.