← 2.1.6 Test inconclusive · runtime-test

Status Line Percentage Fields — runtime test

Hands-on runtime battle-test of Status Line Percentage Fields. Result: INCONCLUSIVE.

Status Line Percentage Fields expose context window consumption as normalized percentages in the Claude Code status line.

What Was Tested

Claude Code 2.1.6 introduced two template variables for the status line: context_window.used_percentage and context_window.remaining_percentage. These complement existing raw token counters by normalizing context to 0-100 scale. The test scenario was to verify that template evaluation produces correct percentages when configured in statusLineTemplate settings.

Test Result: Inconclusive

Status: NOT RUNTIME TESTED. The feature was code-reviewed but not executed in the running environment. The test plan identified the setup requirement (configure statusLineTemplate with the percentage fields) and the success criterion (percentages appear correctly), but did not proceed to observation. No test data exists on whether the template variables resolve to valid numbers, whether they update correctly as context accumulates, or whether they maintain arithmetic consistency (used + remaining = 100).

Why It Matters

Percentages reduce cognitive load compared to raw token accounting. A user facing 2850/40000 tokens used must mentally divide to assess headroom; 7% used is immediate. For agents with tight context margins, at-a-glance percentage display surfaces warnings that might escape raw-token readout. This bridges observable context pressure and operator awareness.

Caveats

No runtime assertions exist. Edge cases remain unmapped: how percentage fields behave when context is exactly full, when timestamp or other updates occur mid-token-count read, or whether they maintain synch across concurrent status-line renders. Template syntax validation and model-specific ranges (Opus vs. Haiku context size differences) were assumed in design but not verified in execution.

Residue: Needs interactive runtime validation with a live Claude Code session running the configured template against actual token-usage scenarios.

Primary source
⎘ 2.1.6/tests/02-status-line-percentages/TEST-RESULTS.mdverbatim from the corpus

Test Results: Status Line Percentage Fields

Feature: Added context_window.used_percentage and context_window.remaining_percentage fields to status line input

Tested: 2026-01-16 Version: 2.1.6 features on 2.1.9

Feature Description

New template variables for status line configuration:

  • context_window.used_percentage - How much of context window is used (0-100)
  • context_window.remaining_percentage - How much remains (0-100)

These complement existing fields like context_window.total_tokens.

Test Result: NOT RUNTIME TESTED

Would need to:

  1. Configure status line template to use these fields
  2. Observe if percentages appear correctly

Expected Usage

In settings.json:

{
  "statusLineTemplate": "{{context_window.used_percentage}}% used | {{context_window.remaining_percentage}}% free"
}

Use Cases

  1. Visual progress - See at a glance how full context is
  2. Warning thresholds - Easier to notice when approaching limits
  3. Simpler than tokens - Percentages more intuitive than raw token counts

Notes

Makes context window monitoring more accessible. Previously had to interpret raw token counts against model limits.

Status: CODE REVIEW (needs config to test)

Evidence & receipt
◇ ed25519 receipt
idtest_d4926f97f34e61d32c25f5d3
alged25519
pubkey9b87705613b1e2fd064d57fa75a6b679d2856ceafad6b1daa8f982493871b6dd
sig71e2aad4cbd505794442e48282bd069a3e90bca200750ebcd8cc660be0b64ad0b48e2fe759cdf253924239e6471623822789116408ca53fd151a9050a55dcc07

Signed with an ed25519 key held off the repo. Anyone can verify against the published public key; nobody without the secret key can forge it. Click verify: it recomputes the signature in your browser. The signature proves integrity and authorship of this exact content — not a third-party timestamp or that the underlying claim is objectively true. signedAt is when the @f3/attest pipeline ran, not when the work happened; the evidence refs carry the source dates.

Connected